Property Price Impacts of Environment-Friendly Transport Accessibility in Chinese Cities 981168832X, 9789811688324

This book seeks to shed light on the role of environment-friendly transport accessibility in determining property prices

130 19 4MB

English Pages 194 [188] Year 2022

Report DMCA / Copyright

DOWNLOAD PDF FILE

Table of contents :
Preface
Acknowledgements
Contents
1 A Short Introduction
1.1 Background
1.1.1 Environment-Friendly Transport
1.1.2 Environment-Friendly Transport Accessibility
1.1.3 Environment-Friendly Transport Accessibility and Property Prices
1.2 Significance
1.3 Structure
References
2 Place-Varying Impact of Metro Accessibility on Property Prices
2.1 Introduction
2.2 Overview of Shenzhen
2.3 Snapshot of the Hedonic Pricing Model
2.4 Property Price Data
2.5 Results
2.5.1 Basic Models: Does URT Accessibility Offer Premiums?
2.5.2 Enhanced Model: Do Transfer Stations Provide Larger Accessibility Benefits Than Regular Stations?
2.5.3 Market Segmentation Analyses: Are URT Accessibility Benefits More Perceptible in the Suburban Area Than in the Urban Area?
2.6 Discussion
2.7 Conclusions
References
3 Time-Varying Impact of Metro Accessibility on Property Prices
3.1 Introduction
3.2 Overview of Hong Kong
3.3 Hypothesis Development
3.4 Snapshot of the Spatial Durbin Model
3.5 Property Price Data
3.6 Results
3.7 Conclusions and Discussion
References
4 Accessibility and Proximity Effects of Bus Rapid Transit on Property Prices
4.1 Introduction
4.2 Overview of Xiamen and Its BRT System
4.2.1 Xiamen
4.2.2 Xiamen BRT System
4.3 Snapshot of Box-Cox Transformation
4.4 Snapshot of Spatial Econometric Models
4.5 Property Price Data
4.6 Results
4.6.1 Basic Models: Do Accessibility and Proximity Effects of BRT Coexist?
4.6.2 Enhanced Models: Are Accessibility-Based Benefits of BRT More Perceptible in the Peripheral Area?
4.7 Policy Implications
4.7.1 Value Capture
4.7.2 Decreasing the Negative Externality of Proximity to the Transit Corridor
4.8 Conclusions
References
5 Accessibility and Proximity Effects of Bus Rapid Transit on Property Prices: Heterogeneity Across Price Quantiles and Space
5.1 Introduction
5.2 Snapshot of the Quantile Regression Model
5.3 Snapshot of the GWR Model
5.4 Property Price Data
5.5 Results
5.5.1 Global Regression
5.5.2 Local Regression
5.6 Discussion
5.7 Conclusions
References
6 Non-linear Relationships Between Bus Rapid Transit and Property Prices
6.1 Introduction
6.2 Snapshot of GBDT
6.3 Dependent and Independent Variables
6.4 Results
6.4.1 Relative Importance of Predictor Variables
6.4.2 Partial Dependence Plots (PDPs) of Predictor Variables
6.4.3 Comparison of GBDT Modeling and Hedonic Regression
6.5 Discussion
6.5.1 Policy Implications
6.5.2 Application of Machine Learning Techniques in the Real Estate Market
6.6 Conclusions
References
7 Accessibility to Bus, by Bus, and Property Prices
7.1 Introduction
7.2 Two-Component Transit Accessibility Metrics
7.3 Property Price Data
7.4 Results
7.4.1 OLS Regression and Box-Cox Transformation
7.4.2 Spatial Regression
7.4.3 Robustness Checks
7.5 Conclusions and Discussion
References
8 Accessibility to Bus, by Bus, and Property Prices: Spatially Varying Relationships
8.1 Introduction
8.2 Property Price Data
8.3 Modeling Results
8.3.1 Global Regression
8.3.2 Local Regression
8.4 Discussion
8.5 Conclusions
References
9 Walking Accessibility to Public Services and Property Prices
9.1 Introduction
9.2 Cumulative Opportunity-Based Accessibility Measure
9.3 Dependent and Independent Variables
9.4 Results
9.4.1 OLS Regression
9.4.2 Box-Cox Transformation
9.4.3 Spatial Regression
9.4.4 Estimation of Walking Accessibility Elasticity
9.4.5 Two-Stage Regression for Robustness Checks
9.5 Conclusions and Discussion
References
10 Access to Basic Public Services Has Higher Shadow Prices in Low-End Properties Than in High-End Properties
10.1 Introduction
10.2 Hypothesis Development
10.3 Snapshot of the Random Forest
10.4 Dependent and Independent Variables
10.5 Results
10.5.1 OLS Regression
10.5.2 Spatial Regression
10.5.3 Random Forest
10.5.4 Summary
10.6 Conclusions and Discussion
10.6.1 Concluding Remarks
10.6.2 Practical Implications
10.6.3 Research Limitations
References
11 Conclusions
11.1 Major Findings
11.2 Policy/Practical Implications
References
Recommend Papers

Property Price Impacts of Environment-Friendly Transport Accessibility in Chinese Cities
 981168832X, 9789811688324

  • 0 0 0
  • Like this paper and download? You can publish your own PDF file online for free in a few minutes! Sign Up
File loading please wait...
Citation preview

Linchuan Yang

Property Price Impacts of Environment-Friendly Transport Accessibility in Chinese Cities

Property Price Impacts of Environment-Friendly Transport Accessibility in Chinese Cities

Linchuan Yang

Property Price Impacts of Environment-Friendly Transport Accessibility in Chinese Cities

Linchuan Yang Department of Urban and Rural Planning School of Architecture Southwest Jiaotong University Chengdu, China

ISBN 978-981-16-8832-4 ISBN 978-981-16-8833-1 (eBook) https://doi.org/10.1007/978-981-16-8833-1 © The Editor(s) (if applicable) and The Author(s), under exclusive license to Springer Nature Singapore Pte Ltd. 2021 This work is subject to copyright. All rights are solely and exclusively licensed by the Publisher, whether the whole or part of the material is concerned, specifically the rights of translation, reprinting, reuse of illustrations, recitation, broadcasting, reproduction on microfilms or in any other physical way, and transmission or information storage and retrieval, electronic adaptation, computer software, or by similar or dissimilar methodology now known or hereafter developed. The use of general descriptive names, registered names, trademarks, service marks, etc. in this publication does not imply, even in the absence of a specific statement, that such names are exempt from the relevant protective laws and regulations and therefore free for general use. The publisher, the authors and the editors are safe to assume that the advice and information in this book are believed to be true and accurate at the date of publication. Neither the publisher nor the authors or the editors give a warranty, expressed or implied, with respect to the material contained herein or for any errors or omissions that may have been made. The publisher remains neutral with regard to jurisdictional claims in published maps and institutional affiliations. This Springer imprint is published by the registered company Springer Nature Singapore Pte Ltd. The registered company address is: 152 Beach Road, #21-01/04 Gateway East, Singapore 189721, Singapore

Preface

Developing an accessible and environment-friendly transport system is an effective means of solving many urban problems, such as traffic congestion and noise/air pollution. Environment-friendly transportation modes (e.g., walking and mass public transport) have gained increasing popularity and attracted substantial attention from, inter alia, policymakers and academics. Environment-friendly transport accessibility should, in theory, affect housing prices, as housing purchasers should be willing to pay extra for such a desirable attribute. As of today, most of the relevant studies have been conducted in the West. There are, however, much fewer empirical studies in China, the largest developing country that is undergoing rapid urbanization and motorization. This book seeks to shed light on the role of environment-friendly transport accessibility (specifically, metro accessibility, bus rapid transit (BRT) accessibility, bus accessibility, and walking accessibility) in determining property prices. Hong Kong, Shenzhen, and Xiamen are chosen as the study area. Hedonic pricing models and their advanced versions (e.g., spatial econometric models) are developed to investigate the relationship between environment-friendly transport accessibility and property prices. The empirical findings are summarized below: First, there are variations in the property price premium stemming from metro accessibility. Specifically, transfer stations provide a larger price premium than regular stations, and the price effect of metro accessibility is more perceptible in the suburban area than in the urban area. Moreover, in a shopping destination, the impact of metro accessibility on retail property prices increases with tourist volume. That is, the implicit price of metro accessibility increases with tourism demand. Second, BRT accessibility (or accessibility to BRT stations) positively affects property prices, while proximity to the BRT corridor has a negative impact on housing prices. In addition, BRT accessibility has a stronger positive effect on property prices in the peripheral area than in the central. Moreover, the heterogeneity across price quantiles and space is determined. For example, the buyers of high-priced properties have a greater willingness to pay for avoiding the nuisances attributed to proximity to the BRT corridor.

v

vi

Preface

Third, bus accessibility can be decomposed into to-bus accessibility and by-bus accessibility. Previous studies generally agree that to-bus accessibility exerts an imperceptible impact on property prices. However, empirical results in this study show that the impact is positive and significant. An explanation is the high quality of bus services in Xiamen that make the to-bus accessibility valuable or desirable. The reliability of bus services makes Xiamen a bus-dependent city. In addition, by-bus accessibility (specifically, bus frequency and travel time to the city center) has a significant positive impact on property prices. Furthermore, significant spatial heterogeneity in the capitalization effects. Last but not least, since a single variable (e.g., Walk Score) cannot comprehensively describe walking accessibility as a property price influencing amenity, this book applies the cumulative opportunity approach to measuring this amenity. In addition, it concludes that walking accessibility has different impacts on housing prices under different conditions. Moreover, access to “basic” public services (e.g., transit services) has a stronger positive impact on the prices of low-end (cheaper) housing than on those of high-end (more expensive) housing. This finding can be explained by the varying demands of wealthy and low-income households. The empirical findings of this study are of practical value including (1) informing urban planners/designers to plan/design cities with an adequate level of environmentfriendly transport accessibility; (2) offering an evidence-based approach to implementing value capture schemes for financing investments in urban infrastructure; and (3) providing the basis for mitigating the negative externality of proximity to the transit corridor, jointly constructing comprehensive hospitals and other compatible amenities, and so forth. Chengdu, China

Linchuan Yang

Acknowledgements

This book reports the research I conducted in recent years. I would like to express my thanks to my colleagues at Southwest Jiaotong University, including but not limited to Prof. Zhongwei Shen, Prof. Xu Cui, Prof. Qing Zhu, Prof. Yang Yu, Dr. Ji Li, Dr. Hongtai Yang, Dr. Qinran Yang, Dr. Xinzi Tang, Dr. Ye Yuan, Dr. Jiao Chen, and Dr. Sining Zhang. Without their tireless encouragement, this book will not appear. I feel deeply indebted to my Ph.D. supervisor, the former Chair Professor and Head of the Department of Real Estate and Construction, The University of Hong Kong (HKU), for his meticulous supervision of my real estate valuation research. His academic intelligence and problem-solving capability greatly inspire me. Moreover, I am grateful to Prof. S. K. Wong (HKU), whose marvelous economic thinking always enlightens me. Also, I truly appreciate the insightful and penetrating comments from Prof. L. W. C. Lai (HKU), Dr. L. H. Choy (HKU), and Dr. Z. Yang (Tsinghua University). Special gratitude goes to the Department of Civil Engineering, HKU (where I finished the M.Phil. program, supervised by Prof. W. Y. Szeto and Dr. R. C. P. Wong) and Xiamen University (where I obtained a bachelor’s degree in engineering and a bachelor’s degree in science). Moreover, my gratitude goes to all my friends. I could still recall vivid moments of us traveling, swimming, hiking, singing, watching movies, playing guitar, savoring cuisines, and reading novels together. All in all, they are part and parcel of my iridescent life. Last but not least, the book is devoted to my parents and wife, who are always around whenever I need help. This book includes a series of my papers published with outstanding collaborators in recent years, which are listed as follows: 1.

Yang, L., Liang, Y., Zhu, Q., & Chu, X. (2021). Machine learning for inference: Using gradient boosting decision tree to assess non-linear effects of bus rapid transit on house prices. Annals of GIS, 27(3), 273–284.

vii

viii

2.

3.

4.

5. 6.

7. 8.

9.

Acknowledgements

Yang, L., Chen, Y., Xu, N., Zhao, R., Chau, K. W., & Hong, S. (2020). Placevarying impacts of urban rail transit on property prices in Shenzhen, China: Insights for value capture. Sustainable Cities and Society, 58, 102140. Yang, L., Chau, K. W., Szeto, W. Y., Cui, X., & Wang, X. (2020). Accessibility to transit, by transit, and property prices: Spatially varying relationships. Transportation Research Part D: Transport and Environment, 85, 102387. Yang, L., Chu, X., Gou, Z., Yang, H., Lu, Y., & Huang, W. (2020). Accessibility and proximity effects of bus rapid transit on housing prices: Heterogeneity across price quantiles and space. Journal of Transport Geography, 88, 102850. Liu, Y., Yang, L., & Chau, K. W. (2020). Impacts of tourism demand on retail property prices in a shopping destination. Sustainability, 12(4), 1361. Yang, L., Chau, K. W., & Chu, X. (2019). Accessibility-based premiums and proximity-induced discounts stemming from bus rapid transit in China: Empirical evidence and policy implications. Sustainable Cities and Society, 48, 101561. Yang, L., Zhou, J., Shyr, O. F., & Huo, D. D. (2019). Does bus accessibility affect property prices? Cities, 84, 56–65. Yang, L., Chau, K. W., & Wang, X. (2019). Are low-end housing purchasers more willing to pay for access to public services? Evidence from China. Research in Transportation Economics, 76, 100734. Yang, L., Wang, B., Zhou, J., & Wang, X. (2018). Walking accessibility and property prices. Transportation Research Part D: Transport and Environment, 62, 551–562.

Chengdu, China

Linchuan Yang

Contents

1

2

A Short Introduction . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 1.1 Background . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 1.1.1 Environment-Friendly Transport . . . . . . . . . . . . . . . . . . . . 1.1.2 Environment-Friendly Transport Accessibility . . . . . . . . 1.1.3 Environment-Friendly Transport Accessibility and Property Prices . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 1.2 Significance . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 1.3 Structure . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . References . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . Place-Varying Impact of Metro Accessibility on Property Prices . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 2.1 Introduction . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 2.2 Overview of Shenzhen . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 2.3 Snapshot of the Hedonic Pricing Model . . . . . . . . . . . . . . . . . . . . . 2.4 Property Price Data . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 2.5 Results . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 2.5.1 Basic Models: Does URT Accessibility Offer Premiums? . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 2.5.2 Enhanced Model: Do Transfer Stations Provide Larger Accessibility Benefits Than Regular Stations? . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 2.5.3 Market Segmentation Analyses: Are URT Accessibility Benefits More Perceptible in the Suburban Area Than in the Urban Area? . . . . . . . . 2.6 Discussion . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 2.7 Conclusions . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . References . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .

1 1 1 2 3 5 6 7 9 9 11 12 13 17 17

19

20 21 23 24

ix

x

3

4

5

Contents

Time-Varying Impact of Metro Accessibility on Property Prices . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 3.1 Introduction . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 3.2 Overview of Hong Kong . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 3.3 Hypothesis Development . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 3.4 Snapshot of the Spatial Durbin Model . . . . . . . . . . . . . . . . . . . . . . . 3.5 Property Price Data . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 3.6 Results . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 3.7 Conclusions and Discussion . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . References . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . Accessibility and Proximity Effects of Bus Rapid Transit on Property Prices . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 4.1 Introduction . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 4.2 Overview of Xiamen and Its BRT System . . . . . . . . . . . . . . . . . . . . 4.2.1 Xiamen . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 4.2.2 Xiamen BRT System . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 4.3 Snapshot of Box-Cox Transformation . . . . . . . . . . . . . . . . . . . . . . . 4.4 Snapshot of Spatial Econometric Models . . . . . . . . . . . . . . . . . . . . 4.5 Property Price Data . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 4.6 Results . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 4.6.1 Basic Models: Do Accessibility and Proximity Effects of BRT Coexist? . . . . . . . . . . . . . . . . . . . . . . . . . . . 4.6.2 Enhanced Models: Are Accessibility-Based Benefits of BRT More Perceptible in the Peripheral Area? . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 4.7 Policy Implications . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 4.7.1 Value Capture . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 4.7.2 Decreasing the Negative Externality of Proximity to the Transit Corridor . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 4.8 Conclusions . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . References . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . Accessibility and Proximity Effects of Bus Rapid Transit on Property Prices: Heterogeneity Across Price Quantiles and Space . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 5.1 Introduction . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 5.2 Snapshot of the Quantile Regression Model . . . . . . . . . . . . . . . . . . 5.3 Snapshot of the GWR Model . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 5.4 Property Price Data . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 5.5 Results . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 5.5.1 Global Regression . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 5.5.2 Local Regression . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 5.6 Discussion . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 5.7 Conclusions . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . References . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .

27 27 28 29 31 32 35 39 40 43 43 45 45 47 49 50 51 52 52

58 61 61 62 63 64

67 67 69 69 71 73 74 79 82 84 85

Contents

xi

6

Non-linear Relationships Between Bus Rapid Transit and Property Prices . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 89 6.1 Introduction . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 89 6.2 Snapshot of GBDT . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 91 6.3 Dependent and Independent Variables . . . . . . . . . . . . . . . . . . . . . . . 91 6.4 Results . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 93 6.4.1 Relative Importance of Predictor Variables . . . . . . . . . . . 93 6.4.2 Partial Dependence Plots (PDPs) of Predictor Variables . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 93 6.4.3 Comparison of GBDT Modeling and Hedonic Regression . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 95 6.5 Discussion . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 97 6.5.1 Policy Implications . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 97 6.5.2 Application of Machine Learning Techniques in the Real Estate Market . . . . . . . . . . . . . . . . . . . . . . . . . . 98 6.6 Conclusions . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 99 References . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 100

7

Accessibility to Bus, by Bus, and Property Prices . . . . . . . . . . . . . . . . . 7.1 Introduction . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 7.2 Two-Component Transit Accessibility Metrics . . . . . . . . . . . . . . . . 7.3 Property Price Data . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 7.4 Results . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 7.4.1 OLS Regression and Box-Cox Transformation . . . . . . . . 7.4.2 Spatial Regression . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 7.4.3 Robustness Checks . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 7.5 Conclusions and Discussion . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . References . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .

103 103 105 106 108 108 110 115 115 119

8

Accessibility to Bus, by Bus, and Property Prices: Spatially Varying Relationships . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 8.1 Introduction . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 8.2 Property Price Data . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 8.3 Modeling Results . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 8.3.1 Global Regression . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 8.3.2 Local Regression . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 8.4 Discussion . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 8.5 Conclusions . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . References . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .

123 123 125 128 128 132 134 137 139

Walking Accessibility to Public Services and Property Prices . . . . . . 9.1 Introduction . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 9.2 Cumulative Opportunity-Based Accessibility Measure . . . . . . . . . 9.3 Dependent and Independent Variables . . . . . . . . . . . . . . . . . . . . . . .

141 141 143 145

9

xii

Contents

9.4

Results . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 9.4.1 OLS Regression . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 9.4.2 Box-Cox Transformation . . . . . . . . . . . . . . . . . . . . . . . . . . 9.4.3 Spatial Regression . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 9.4.4 Estimation of Walking Accessibility Elasticity . . . . . . . . 9.4.5 Two-Stage Regression for Robustness Checks . . . . . . . . . 9.5 Conclusions and Discussion . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . References . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .

145 145 148 148 153 154 154 157

10 Access to Basic Public Services Has Higher Shadow Prices in Low-End Properties Than in High-End Properties . . . . . . . . . . . . . 10.1 Introduction . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 10.2 Hypothesis Development . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 10.3 Snapshot of the Random Forest . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 10.4 Dependent and Independent Variables . . . . . . . . . . . . . . . . . . . . . . . 10.5 Results . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 10.5.1 OLS Regression . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 10.5.2 Spatial Regression . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 10.5.3 Random Forest . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 10.5.4 Summary . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 10.6 Conclusions and Discussion . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 10.6.1 Concluding Remarks . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 10.6.2 Practical Implications . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 10.6.3 Research Limitations . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . References . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .

159 159 160 161 162 167 167 167 171 171 173 173 173 174 175

11 Conclusions . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 11.1 Major Findings . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 11.2 Policy/Practical Implications . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . References . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .

179 179 181 182

Chapter 1

A Short Introduction

1.1 Background 1.1.1 Environment-Friendly Transport At present, numerous cities worldwide grapple with various conspicuous problems, such as traffic congestion, air/noise pollution, environmental degradation, limited green open space, carbon dioxide/greenhouse gas emissions, and diminishing ecological amenities (Ewing & Cervero, 2010), and sustainability issues have widely been presented in today’s cities (Bao & Lu 2020; Bao et al., 2019). With unprecedentedly rapid economic growth and the acceleration of urbanization, China is far from an exception in this regard. Against such a background, reducing the motorized trip frequency, uplifting the proportion of non-motorized (mainly comprised by walking and cycling) trips, decreasing vehicle travel distances, and increasing vehicle occupancy levels (e.g., transit) (Cervero & Kockelman, 1997) have been frequently regarded as critical transportation objectives. Currently, establishing an accessible and environment-friendly transport system is deemed an effective approach to solving many urban problems (Cervero & Kockelman, 1997). Put in another way, environment-friendly transport means (or modes), including but not limited to walking and public transit, have gained immense popularity. For example, in China, since 2011, the Ministry of Transport of China has initiated to develop many cities into transit metropolises (gongjiao dushi); and in 2012, the Ministry of Transport, the Ministry of Finance, and the National Development and Reform Commission jointly released Guidance on Strengthening the Development of Walking and Cycling Systems (guanyu jiaqiang chengshi buxing he zixingche jiaotong xitong jianshe de zhidao yijian). Even the United States, which has long been infamous for its car dependence, is witnessing a resurgence of public transit systems. As a low-carbon, environment-friendly, and sustainable non-motorized travel mode and the most prevailing aerobic exercise, walking has garnered substantial scholarly attention in recent years. It is easily incorporated into a daily routine and © The Author(s), under exclusive license to Springer Nature Singapore Pte Ltd. 2021 L. Yang, Property Price Impacts of Environment-Friendly Transport Accessibility in Chinese Cities, https://doi.org/10.1007/978-981-16-8833-1_1

1

2

1 A Short Introduction

has a wide array of environmental, health, and social benefits. First, the environmental benefits of walking include decreasing car use, air pollution, and greenhouse gas emissions (Heath et al., 2012; Woodcock et al., 2009). Second, walking requires the physical activity of human beings and helps beat the notorious sedentary and inactive lifestyle. It, therefore, has a wide array of known health benefits, such as a decreased risk of depression and obesity (Woodcock et al., 2009). The health benefits are relevant to improvements in quality of life and subjective wellbeing (Bird et al., 2013). Last, aside from well-recognized environmental and health benefits, walking also has quite a few social benefits, such as increasing social capital, encouraging social interaction, fostering weak social ties, facilitating social and political participation, decreasing societal costs, and boosting community cohesion, trust, and livability (Leyden, 2003; Lund, 2002). All in all, walking is beneficial to pedestrians, the community, and society. However, walking is a way of physical activity and thus cannot meet all demands for travel, especially those for medium- and long-distance travel. As an alternative to the car, transit (or public transport) is advocated by numerous scholars to meet medium- and long-distance travel demands. As an economic-feasible, lowcarbon, and environment-friendly travel mode, transit has been widely seen to have the potential to reduce car dependence and to facilitate high-density and mixed landuse development patterns (Cervero & Kockelman, 1997). Accordingly, developing an attractive, accessible, and affordable public transport system that meets the needs of transit users is imperative for sustainable development. Evidently, public transport is suitable for high-density and compact cities, particularly those in developing states.

1.1.2 Environment-Friendly Transport Accessibility Transport accessibility (“accessibility” for shorthand hereafter) is an extensively studied concept in various fields, including transportation, urban planning, and geography. It is usually regarded as the “potential of opportunities for interaction” (Hansen, 1959, p. 73) and is essential for transport mode choice and usage. However, accessibility lacks a unified and unambiguous definition, which often hinges on the problem and context (Kwan, 1998). Selected definitions are summarized in Table 1.1. Notably, accessibility is thought to be influenced “by the qualities of the transport system (reflecting the travel time or the costs of reaching a destination) on the one hand and by the qualities of the land-use system (reflecting the qualities of potential destinations), on the other hand” (Straatemeier, 2008, p. 128). In addition to the diversity of definitions, accessibility measures are also rather diverse. They include but are not limited to travel impedance (travel distance/time and monetary cost), cumulative opportunity-based measures, gravity-based measures, utility-based measures, and constraints-based measures. Walking accessibility can be defined as the ease of reaching essential destinations in the walk-in catchment area. This definition is analogous to that provided by Pivo

1.1 Background

3

Table 1.1 Selected definitions of transport accessibility Definition

Sources

Potential of opportunities for interaction

Hansen (1959, p. 73)

Ease with which any land-use activity can be reached from a location using a particular transport system

Burns and Golob (1976, p. 175)

Freedom of individuals to decide whether or not to participate in different activities

Burns (1979, p. 1)

Ease with which activities may be reached from a given location using a particular transportation system

Morris et al. (1979, p. 92)

Ease of reaching places

Cervero (1996, p. 1)

Number of activities which can be reached from a certain location

Geurs and Ritsema van Eck (2001, p. 19)

Extent to which land-use and transport systems enable (groups of) individuals to reach activities or destinations by means of a (combination of) transport mode(s)

Geurs and van Wee (2004, p. 128)

Ease of reaching valued destinations

Sun et al. (2017, p. 442)

and Fisher (2011, p. 187), namely “the degree to which an area with walking distance of a property encourages walking trips from the property to other destinations.” For the sake of life convenience, enhancing walking accessibility to amenities (e.g., shopping centers and schools) is generally desirable. Moreover, an inherent drawback of transit systems is that they only provide stopto-stop services instead of door-to-door services. Thus, access to the stop or station affects residents’ travel mode choice. If the distance to the transit stop or station exceeds a certain threshold, then residents may not consider traveling by transit. That is, people would often turn to private cars (as long as they can afford them) when decent transit services are inaccessible. Therefore, transit accessibility is often interpreted as the ease of using transit or access to transit (to-transit accessibility). Interestingly, in a handful of past studies, transit accessibility is considered the combination of to-transit accessibility and by-transit accessibility (i.e., the convenience of accomplishing specific activities by transit).

1.1.3 Environment-Friendly Transport Accessibility and Property Prices Characterized by structural inflexibility, temporal durability, as well as spatial fixity, a property (e.g., house) is a commodity with a series of attributes (or characteristics), such as floor area, transit accessibility, and proximity to the city center (So et al., 1997). After marketization, the house in urban China has become a commodity

4

1 A Short Introduction

distinguished by a variety of attributes (e.g., size and distance to the city center), many of which have been claimed to affect its price. As an observable utility-generating (or utility-bearing) attribute, environment-friendly transport accessibility (i.e., walking accessibility and transit accessibility) presumably affects property prices. Residents often desire some degree of environment-friendly transport accessibility and thus may be willing to pay high prices for it. This is expected to be translated to price premiums. The theoretical underpinnings of the association between transit and property (or land) prices (or rents) can be traced back to land rent theory (Alonso, 1964; Mills, 1967; Muth, 1969). An enriching understanding of the relationship between environment-friendly transport accessibility and property prices is crucial as it is of policy and practical relevance, beneficial to decision making. For example, it is widely recognized that constructing and maintaining huge transit projects, such as the metro, needs uncountable money, which often encumbers local governments with debt. An essential approach to developing such projects without increasing government debt burden is searching for new financial models. One credible model is the value capture scheme. Typically, such huge transit projects are expected to increase property values and create economic benefits, and the value increment can fund existing/new projects. Theoretically, provided that value capture schemes are in place, these projects can be developed without government subsidy. Understanding the relationship between transit accessibility and property prices is the first and foremost step in implementing value capture schemes for transit project financing. Numerous empirical studies have focused on the effects of environment-friendly transport accessibility, especially transit accessibility, on property prices. Among transit modes, rail (including high-speed rail, commuter rail, and light/heavy rail) and bus rapid transit (BRT) systems have received much attention in a burgeoning number of studies. The majority of previous property valuation studies have reached a consistent conclusion that to-transit accessibility enhances nearby property prices. There are several meta-analyses and literature reviews on rail and/or BRT systems and property prices, including Debrezion et al. (2007), Mohammad et al. (2013), Stokenberga (2014), Higgins and Kanaroglou (2016), and Ingvardson and Nielsen (2018). They suggested: (1) rail and BRT systems likely enhance nearby property prices, but their price effect estimates considerably vary across previous empirical studies; (2) the price effect is influenced by a host of factors, including but not limited to station-area land use planning, accessibility to stations, the type of transit service and property, demographics, and empirical modeling methods; (3) quintessentially, the rail has a higher effect on nearby property prices than BRT; and (4) the price effects of rail and BRT systems vary across property types (e.g., residential, office, retail, and industrial). Unlike rail and BRT systems, there is a paucity of research looking at the bus and examining the association between its accessibility and property prices. In stark contrast to the price effects of fixed-guideway transit modes that are relatively consistent, the capitalization effects of the bus are highly elusive: mixed and even conflicting results exist. Most research has discovered that the price effects of bus accessibility are imperceptible. Cervero and Kang (2011, p. 103) stated that “traditional bus transit

1.1 Background

5

services … fail to confer appreciable accessibility benefits.” Empirically, based on hundreds of transactions of single-family properties in Denver between 1969 and 1974, Koutsopoulos (1977) found no appreciable effects of bus accessibility for most routes. So et al. (1997) determined that the price effects of bus accessibility are insignificant in Hong Kong, a city with high metro and bus ridership. Moreover, several studies have identified the negative price effects of bus accessibility, which may be attributable to its nuisances (e.g., congestion and noise/air pollution). Cao and Hough (2008) found that bus accessibility negatively influences property rentals in Fargo, the United States, on the basis of hedonic modeling. Furthermore, few studies have determined positive bus accessibility effects on property prices. Zheng and Kahn (2008) concluded that bus accessibility is positively associated with property prices in Beijing, China.

1.2 Significance Predominately as a response to various conspicuous urban problems, environmentfriendly transport has aroused substantial attention from governments, researchers, and so forth. Environment-friendly transport accessibility presumably affects property prices, as housing purchasers may have a willingness to pay extra for such a largely favorable and desirable attribute. However, the vast majority of existing studies are conducted in the West. Comparatively, China, which is undergoing rapid urbanization and motorization, has received much less scholarly attention and should ignite a tremendous fascination in local and international researchers alike. In this context, severe concerns over walkability and transit use are extant as private cars encroach sidewalks and open spaces rapidly and excessively and compete with transit furiously. Furthermore, a large number of China-specific characteristics (e.g., preferences for proximity to essential service, short commuting, and high population density) make walkability and transit use more significant (Cao, 2014). Due to the differences of contexts in many aspects (e.g., car dependence versus transit dependence), conclusions drawn in the West cannot be generalized to urban China. In light of the above, this book puts effort in shedding light on the relationship between environment-friendly transport (specifically, metro, BRT, bus, and walking) accessibility and property prices in urban China, where few empirical studies have been conducted on quantifying the price premiums attributable to environment-friendly accessibility to date and where many cities are rapidly growing and expanding at an unprecedented rate. Based on extant studies and related theories, this book formulates many testable or refutable hypotheses on environment-friendly transport accessibility (including metro accessibility, BRT accessibility, bus accessibility, and walking accessibility) and property prices. This book conducts a few empirical studies in several Chinese cities (i.e., Hong Kong, Shenzhen, and Xiamen) and investigates the relationship between environment-friendly transport accessibility and property prices.

6

1 A Short Introduction

Bear in mind that some hypotheses were first proposed or empirically tested in similar contexts. Moreover, the book reaches some conclusions that are in sharp contrast with those of existing studies in the West but highly reasonable in the Chinese context (and other settings with similar urban characteristics). Reasons for the discrepancy are also proposed. The empirical findings of this book enhance our understanding of the role of environment-friendly transport in determining property prices. Some findings, such as significant metro accessibility premiums (e.g., Chap. 2), are consistent with previous studies. However, some have seldom been pointed out previously, for example, price discounts stemming from proximity to the BRT corridor (e.g., Chaps. 4 and 5) and non-linear associations between property attributes and property prices (see Chap. 6). Others, for example, a larger willingness to pay for access to basic public services for low-end property purchasers (e.g., Chap. 10), have not been proposed in the existing research. As such, the book dramatically supplements the extant literature. More importantly and interestingly, as the author argued before, due to contextual differences, some findings, such as significant to-bus accessibility premiums (e.g., Chaps. 7 and 8), cannot be generalized to western settings. But the author suspects that they could apply to other cities with a similar transport system. Considering all the above, this book is academically justified. Moreover, the book has profound practical implications, such as value capture schemes, the mitigation of the negative externality of proximity to the transit corridor, and joint construction of comprehensive hospitals with other compatible amenities. All in all, the book can serve as a valuable reference for researchers who work on similar topics and institutions and governments that concern the relationship between transport accessibility and property prices and/or intend to implement policy measures for value capture, reduction in the negative externality of transit, etc.

1.3 Structure This book consists of eleven chapters, and the remainder is organized as follows: Chapters 2 and 3 focus on place-varying and time-varying, respectively, associations between metro accessibility and property prices. Chapters 4, 5, and 6 investigate the linear or non-linear relationship between accessibility to BRT stations and proximity to the BRT corridor and property prices and examine the heterogeneity across price quantiles and space. Chapters 7 and 8 identify the association between accessibility to bus, by bus, and property prices and determine the spatial heterogeneity in the relationship. Chapter 9 scrutinizes the effect of walking accessibility to various public services, which is measured by the cumulative opportunity approach, on property prices. Chapter 10 compares the shadow prices of access to basic public services for low- and high-end properties and reveals that the access has a higher price for lowend properties than for high-end properties. Chapter 11 summarizes and discusses the findings.

1.3 Structure

7

With regard to the study area, Chaps. 2 and 3 record studies conducted in Shenzhen and Hong Kong (two international cities located in the Guangdong–Hong Kong– Macao Greater Bay Area), respectively. All the following chapters document studies based on Xiamen, a city situated on the southeastern coast of China and at the heart of the Western Taiwan Straits Economic Zone.

References Alonso, W. (1964). Location and land use: Toward a general theory of land rent. Harvard University Press. Bao, Z., Lu, W., Chi, B., Yuan, H., & Hao, J. (2019). Procurement innovation for a circular economy of construction and demolition waste: Lessons learnt from Suzhou, China. Waste Management, 99, 12–21. Bao, Z., & Lu, W. (2020). Developing efficient circularity for construction and demolition waste management in fast emerging economies: Lessons learned from Shenzhen, China. Science of the Total Environment, 724, 138264. Bird, E. L., Baker, G., Mutrie, N., Ogilvie, D., Sahlqvist, S., & Powell, J. (2013). Behavior change techniques used to promote walking and cycling: A systematic review. Health Psychology, 32(8), 829–838. Burns, L. D., & Golob, T. F. (1976). The role of accessibility in basic transportation choice behavior. Transportation, 5(2), 175–198. Burns, L. D. (1979). Transportation. Lexington Books. Cao, X. J. (2014). Residential self-selection in the relationships between the built environment and travel behavior: Introduction to the special issue. Journal of Transport and Land Use, 7(3), 1–3. Cao, X. J., & Hough, J. A. (2008). Hedonic value of transit accessibility: An empirical analysis in a small urban area. Journal of the Transportation Research Forum, 47(3), 171–183. Cervero, R. (1996). Paradigm shift: From automobility to accessibility planning. University of California. Cervero, R., & Kang, C. D. (2011). Bus rapid transit impacts on land uses and land values in Seoul, Korea. Transport Policy, 18(1), 102–116. Cervero, R., & Kockelman, K. (1997). Travel demand and the 3Ds: Density, diversity, and design. Transportation Research Part D: Transport and Environment, 2(3), 199–219. Debrezion, G., Pels, E., & Rietveld, P. (2007). The impact of railway stations on residential and commercial property value: A meta-analysis. Journal of Real Estate Finance and Economics, 35(2), 161–180. Ewing, R., & Cervero, R. (2010). Travel and the built environment: A meta-analysis. Journal of the American Planning Association, 76(3), 265–294. Geurs, K. T., & Ritsema van Eck, J. R. (2001). Accessibility measures: Review and applications. National Institute of Public Health and the Environment. Geurs, K. T., & van Wee, B. (2004). Accessibility evaluation of land-use and transport strategies: Review and research directions. Journal of Transport Geography, 12(2), 127–140. Hansen, W. G. (1959). How accessibility shapes land use. Journal of the American Institute of Planners, 25(2), 73–76. Heath, G. W., Parra, D. C., Sarmiento, O. L., Andersen, L. B., Owen, N., Goenka, S., Montes, F., & Brownson, R. C. (2012). Evidence-based intervention in physical activity: Lessons from around the world. The Lancet, 380(9838), 272–281. Higgins, C. D., & Kanaroglou, P. S. (2016). Forty years of modelling rapid transit’s land value uplift in North America: Moving beyond the tip of the iceberg. Transport Reviews, 36(5), 610–634. Ingvardson, J. B., & Nielsen, O. A. (2018). Effects of new bus and rail rapid transit systems—An international review. Transport Reviews, 38(1), 96–116.

8

1 A Short Introduction

Koutsopoulos, K. C. (1977). The impact of mass transit on residential-property values. Annals of the Association of American Geographers, 67(4), 564–576. Kwan, M. P. (1998). Space-time and integral measures of individual accessibility: A comparative analysis using a point-based framework. Geographical Analysis, 30(3), 191–216. Leyden, K. M. (2003). Social capital and the built environment: The importance of walkable neighborhoods. American Journal of Public Health, 93(9), 1546–1551. Lund, H. (2002). Pedestrian environments and sense of community. Journal of Planning Education and Research, 21(3), 301–312. Mills, E. S. (1967). An aggregative model of resource allocation in a metropolitan area. The American Economic Review, 57(2), 197–210. Mohammad, S. I., Graham, D. J., Melo, P. C., & Anderson, R. J. (2013). A meta-analysis of the impact of rail projects on land and property values. Transportation Research Part A: Policy and Practice, 50, 158–170. Morris, J. M., Dumble, P. L., & Wigan, M. R. (1979). Accessibility indicators for transport planning. Transportation Research Part A: General, 13(2), 91–109. Muth, R. F. (1969). Cities and housing: The spatial pattern of urban residential land use, third series: Studies in business and society. University of Chicago Press. Pivo, G., & Fisher, J. (2011). The walkability premium in commercial real estate investments. Real Estate Economics, 39(2), 185–219. So, H. M., Tse, R. Y., & Ganesan, S. (1997). Estimating the influence of transport on house prices: Evidence from Hong Kong. Journal of Property Valuation and Investment, 15(1), 40–47. Stokenberga, A. (2014). Does bus rapid transit influence urban land development and property values: A review of the literature. Transport Reviews, 34(3), 276–296. Straatemeier, T. (2008). How to plan for regional accessibility? Transport Policy, 15(2), 127–137. Sun, B., Ermagun, A., & Dan, B. (2017). Built environmental impacts on commuting mode choice and distance: Evidence from Shanghai. Transportation Research Part D: Transport and Environment, 52, 441–453. Woodcock, J., Edwards, P., Tonne, C., Armstrong, B. G., Ashiru, O., Banister, D., Beevers, S., Chalabi, Z., Chowdhury, Z., Cohen, A., & Franco, O. H. (2009). Public health benefits of strategies to reduce greenhouse-gas emissions: Urban land transport. The Lancet, 374(9705), 1930–1943. Zheng, S., & Kahn, M. E. (2008). Land and residential property markets in a booming economy: New evidence from Beijing. Journal of Urban Economics, 63(2), 743–757.

Chapter 2

Place-Varying Impact of Metro Accessibility on Property Prices

2.1 Introduction Urban rail transit (URT) is an important mode of public transport, which is characterized by high speed, high efficiency, safety, as well as large capacity (Lin & Du, 2017). Based on these uniformly favorable and desirable characteristics, URT claims the top priority in many big cities around the world. Moreover, URT is reportedly the dominant transportation mode in various metropolitan cities worldwide. For example, in Tokyo, Seoul, and Hong Kong, URT constitutes 48%, 38%, and 30%, respectively, of trips made by residents. In recent years, the spread of URT gains momentum in developing countries such as China. In 2017, 34 Mainland Chinese cities owned URT systems, totaling 165 lines and 5033 km of mileage in terms of operation. A total of 50 Mainland Chinese cities are scheduled to operate a URT system in 2020 according to the official plan of the National Development and Reform Commission of China (Three-year Action Plan for Major Projects in Transportation Infrastructure Construction). Nevertheless, from construction to maintenance, URT systems incur huge capital costs. This problem is especially apparent among city governments with limited budgets. In some cities, URT projects are disapproved, if not suspended, to prevent fiscal distress. For example, in November 2017, the central government of China released guidelines to put a check on excessive URT construction. With such stringent rules, a few URT projects have been halted during construction (Sun et al., 2017). An essential step in implementing value capture schemes is to understand how URT projects raise land/property prices. Existing studies on URT’s impact on land/property prices abound, albeit mainly focusing on developed countries, such as the US, Australia, and the Netherlands. Comparatively, limited research has focused on developing countries. Moreover, the practice of value capture for URT development is far from mature in both China and many other developing countries (Sharma & Newman, 2018). On the other hand, many cities in China are aiming to 26 operate URT by 2020. This shows that rapid urbanization in China has stimulated the demand for URT. Thus, developing URT projects using value capture schemes is © The Author(s), under exclusive license to Springer Nature Singapore Pte Ltd. 2021 L. Yang, Property Price Impacts of Environment-Friendly Transport Accessibility in Chinese Cities, https://doi.org/10.1007/978-981-16-8833-1_2

9

10

2 Place-Varying Impact of Metro Accessibility on Property Prices

an imperative priority. Furthermore, unlike the US URT systems with high subsidies from the federal government, URT construction and operation in China generally do not obtain any funding from the central and provincial governments (Yang et al., 2016). As such, municipal governments in China have strong economic incentives to utilize feasible and effective URT financial models. Feasible value capture schemes are ideal for sustaining China’s URT development. Accordingly, the relationship between URT accessibility and property prices in China should be examined before crafting value capture schemes. Numerous empirical studies have examined the effects of URT on land/property values, but they produced rather mixed, even conflicting results. One of the most comprehensive research is the work of Bowes and Ihlanfeldt (2001), which suggested that the holistic price effects of URT are a tradeoff between positive (e.g., commuting cost saving) and negative effects (e.g., noise or air pollution). Bowes and Ihlanfeldt (2001) further summarized that the price effects of URT originate from 4 aspects, namely reducing commuting costs, increasing neighborhood retail activity, providing nuisances, and bringing about more crime. A large body of extant literature has used sophisticated econometric models to investigate the capitalization effects of URT accessibility on land/property prices. It suggests that URT enhances nearby land/property prices. However, most, though not all, previous studies are silent on (1) whether or not transfer stations provide larger property price premiums than non-transfer (or regular) stations, and (2) whether or not the benefits stemming from URT accessibility are more perceptible in the suburban area than in the urban area. A profound and thorough understanding of the differences is essential for diversified (or differential, location-specific) value capture schemes. That said, with a good understanding of the different capitalization effects, evidence-based diversified value capture schemes can be crafted for different areas within a city. Based on 722 residential complexes within the distance of 2 km from Line 5 in Shenzhen, China, we initially estimate basic hedonic pricing models to determine the correlations between URT accessibility and property prices. Then, to evaluate the contribution of transfer stations, we introduce an additional explanatory variable and develop an enhanced model. Finally, we divide the sample into two groups according to location, namely the urban and suburban areas, and calibrate separate models for the two groups. The results of the Chow test demonstrate that different behaviors can be observed in the two sub-markets. Notably, we use both continuous and dummy variables to capture URT accessibility. The contributions of this study are as follows: (1) determining the capitalization effects of URT accessibility in a Chinese city; (2) quantifying and comparing the different price effects of URT transfer and regular stations; (3) evaluating whether or not the price effects of URT accessibility are more perceptible in the suburban area than in the urban area; and (4) providing insights for designing value capture schemes for URT financing. The remainder of this chapter is displayed as follows. Section 2.2 offers an overview of the study area, Shenzhen. Section 2.3 presents the property price data utilized. Section 2.4 succinctly introduces the hedonic pricing model and hedonic variables. Section 2.5 shows the modeling results. Section 2.6 discusses the findings. Section 2.7 concludes the chapter and points out directions for forthcoming research.

2.2 Overview of Shenzhen

11

2.2 Overview of Shenzhen Shenzhen is a major coastal city in Guangdong province, China. It covered 1997.27 km2 and had a permanent population of 12.52 million in 2017. Shenzhen is the first and most successful special economic zone (SEZ) in China and has been a touchstone for the reform and opening-up as well as an experimental ground for the practice of market capitalism. (The Shenzhen SEZ originally consisted of Luohu, Futian, Yantian, and Nanshan districts and expanded to the whole city in July 2010) Shenzhen has transformed from a remote, tiny fishing village to a first-tier city in Mainland China as well as a famous, prosperous international metropolis with an annual GDP of 2.24 trillion yuan (US$1 = 6.85 yuan) in 2017. The city consists of nine districts and one new area. Shenzhen is the sixth city to provide URT services in Mainland China, following Beijing, Tianjin, Shanghai, Guangzhou, and Wuhan. It opened its first URT line on December 28, 2004. At present, the Shenzhen URT has a total of 8 lines, 166 stations, and 286 km of mileage in operation (Fig. 2.1). The eight URT lines connect six districts (Futian, Luohu, Nanshan, Bao’an, Longgang, and Longhua districts). The average daily passenger volume was approximately 2.99 million, and the annual passenger volume was 1.297 million in 2016. URT accounts for 1/4 trips of city dwellers, which illustrates its critical role in the urban transportation system (Yang et al., 2016). Shenzhen is an ideal place to explore the relationship between URT and property values for the following four reasons. First, the URT system of Shenzhen opened in 2004 and is currently well developed. It constitutes a high proportion of trips made by the residents. Second, influenced by the reform and opening-up policy, Shenzhen is divided into two parts, the area within the original SEZ (guannei) and the area out of it (guanwai). The two areas show substantial differences in many aspects, such as infrastructure construction, land use, and the built environment. This feature is

Fig. 2.1 Map of Shenzhen metro

12

2 Place-Varying Impact of Metro Accessibility on Property Prices

applicable to examine the sub-market effect. Third, Shenzhen acted as a pioneer in actively adopting the “rail plus property (R + P)” program in Mainland China and was designated as a transit metropolis (gongjiao dushi) by the Ministry of Transport of China in 2011. These features jointly assure that the URT system can be developed and operated long-lastingly through constantly learning lessons from Hong Kong (Yang et al., 2016) and receiving extra fundings from high-level governments. Last, the development pattern of the city and that of URT have been widely considered to be a critical paradigm for numerous Mainland Chinese cities. The development of this city has become the focus of society and attracted various spotlights from home and abroad.

2.3 Snapshot of the Hedonic Pricing Model Assuming that implicit prices for property characteristics can be reflected by viewing residents’ willingness to pay, the hedonic price model is a popular revealedpreference method used to explain the price of a good (or commodity, product) in terms of its own characteristics, from which utility is derived. Two pioneering scholars made significant contributions to the development of the hedonic price model. Lancaster (1966) developed a theory of consumer demand, defined utility using the attributes of the goods, and specified that what consumers want to obtain is not goods themselves but the attributes of the goods. He offered microeconomic foundations for analyzing utility-generating (or utility-bearing) attributes and provided the theoretical basis of the hedonic price model. Additionally, Rosen (1974) formulated a seminal theory of hedonic prices and unfolded the myth of how markets work for heterogeneous goods. Typically, the hedonic price model regresses property price/rental (Y, dependent variable) against a bundle of utility-bearing property attributes (Xs, independent variables). Normally, independent variables of the hedonic price model fall into three categories: structural (e.g., gross floor area), location (e.g., proximity to the city center), and neighborhood (e.g., school district and sea view). The lists of popular hedonic variables can be found in Chau and Chin (2003), Malpezzi et al. (2003), Sirmans et al. (2005), and Ko and Cao (2013). Typically, the hedonic pricing model regresses property prices against a host of utility-bearing characteristics of properties. The hedonic pricing model can be written as follows:  βi X i + ε, P = β0 + i

where P is the property price; β0 is a constant; X i is the ith property characteristic (e.g., size and distance to the city center); βi is the coefficient of X i ; and ε is an error term. β0 and βi are parameters to be concurrently estimated. Refer to Chau and Chin (2003), Malpezzi (2003), Bartholomew and Ewing (2011), Krause and Bitter (2012),

2.3 Snapshot of the Hedonic Pricing Model

13

and Sirmans et al. (2005) for detailed reviews of the literature. Besides evaluating shadow prices of property characteristics, hedonic price models can be used in the real estate market in other ways, such as constructing real estate price indices, providing automated valuations of properties, and testing for market segmentation (Hill, 2013). The hedonic pricing model has several basic functional forms, such as linear, semi-log, and double-log forms. The linear function is usually avoided because the assumption of constant marginal implicit prices is not tenable in most, if not all, cases. Comparatively, non-linear models are generally preferred.

2.4 Property Price Data In the empirical analysis, focusing on a small area (instead of the whole city) helps properly control for confounders (or confounding factors) in modeling property prices (Chau & Ng, 1998; Yang et al., 2018, 2019, 2020a, 2020b). In this study, Line 5 (Huanzhong Line) is selected as the investigation URT line (Fig. 2.1 and Table 2.1) for the following reasons. First, this line began to construct in May 2009 and opened in June 2011. In operation for over six years, the line has greatly stimulated the development of surrounding real estate markets. This provides sufficient samples for empirical analysis. Second, Line 5 intersects with seven other lines and has 27 stations (consisting of 6 transfer and 21 regular stations). This feature enables us to compare the different price effects of transfer and regular stations. Last, Line 5 runs from Qianhaiwan to Huangbeiling and connects the urban area (Luohu and Nanshan districts) and the suburban area (Bao’an and Longgang districts) (see Fig. 2.1). This feature enables us to compare the price effects of stations in the urban and suburban areas. Table 2.1 Classifications of URT stations on Line 5

Classification

District

Number of stations

Total

Central area

Nanshan

6

10

Luohu

4

Bao’an

9

Longgang

8

Bao’an

2

Longgang

1

Nanshan

2

Luohu

1

Bao’an

7

Longgang

7

Nanshan

4

Luohu

3

Suburban area Transfer station

Regular station

17 6

21

14

2 Place-Varying Impact of Metro Accessibility on Property Prices

A shortcoming of hedonic pricing models (detailed in Sect. 2.4) is the missing variables bias. Using aggregate (or residential-complex-level, housing-project-level) data rather than disaggregate (or property-level data) would relieve the missing variables bias because it is more difficult to control for the quality heterogeneity (e.g., landscape view) of property observations than that of residential complex observations. Hence, the residential complex is chosen as the basic sample unit here, and resale (or second-hand) residential complexes located within the distance of 2 km of URT stations on Line 5 are sampled. The asking price and location data of a total of 722 residential complexes were obtained in late November 2017 from sz.fang.com (known as Sou-fang.com before), a major real estate agency website in Mainland China. Among them, there are 325 complexes located within the original SEZ (the urban area of the city), with 397 outside of the area. The reasons for confining our sample to resale properties (instead of newly built properties) are as follows. First, the prices of resale properties are more stable and have lived through longer influences from URT stations than newly built properties. Second, the prices of newly-built properties are much more easily influenced by actions carried out by governments (e.g., the home-purchase restriction and credit restriction) than those of resale properties. Lastly and perhaps most importantly, newly built residential complexes in Shenzhen, especially its urban area, are now limited and fairly dispersed. This feature makes statistical modeling very difficult because the sample size is highly limited, and confounders are not easily controlled for at the acceptable level. To reflect the relationship between URT accessibility and property prices accurately, distance-based URT accessibility variables are measured by two types of variables, namely continuous and dummy variables. First, the continuous variable is used. To access whether an inverted-U pattern exists between URT accessibility and property prices, the hedonic pricing model in this study is written as follows: ln P = β0 + α1 lnDMETRO + α2 lnDMETRO2 +



βi X i + ε,

i

where lnP is the natural log of the average price of the residential complex; DMETRO is the distance to the nearest URT station; α1 and α2 are the coefficients of lnDMETRO (the natural log of DMETRO) and lnDMETRO2 ; and other variables are defined as above. Alternatively, dummy variables that reflect URT accessibility are used. As such, the hedonic pricing model is written as follows: ln P = β0 +

 n

γn DMETRO_n +



βi X i + ε,

i

where DMETRO_n is the nth dummy variable, which equals one if DMETRO is within the nth range and zero otherwise; γn is the coefficient of DMETRO_n; and other variables are defined as above.

2.4 Property Price Data

15

Table 2.2 Definitions, quantifications, and expected signs of variables Classification

Variable (code)

Description

Expected sign

Reference

Dependent variable

Price (P)

Average price of the residential complex (104 yuan/m2 )

Control variable

Age (AGE)

Age of the residential complex in years

Property Management Fee (FEE)

Money paid monthly + to provide the property management enterprise that offers a full menu of services (e.g., sanitation and virescence management) to residents, reflecting the quality of the property management service (yuan/m2 per month)

Wen et al. (2014, 2019)

Floor area ratio (FAR)

Ratio of total floor − areas to the site area of the residential complex

Sirmans et al. (2005), Zheng and Kahn (2008), Tsai et al. (2016)

Green coverage rate (GREEN)

Ratio of total green land areas to the site area of the residential complex

+

Tsai et al. (2016)

Parking ratio (PARK)

Number of parking spots per household

+

Chau and Chin (2003)

Distance to CBD (DCBD)

Network distance (or real-life distance) to the CBD of Shenzhen (m)



Chau and Ng (1998), Chau and Chin (2003), Yang et al. (2019), Zheng and Kahn (2013)

Education quality (EDU)

Quality of + neighborhood education facilities, graded by sz.fang.com, ranging from 1 (lowest) to 5 (highest)

Chau and Ng (1998), Chau and Chin (2003), Yang et al. (2019)

Business environment (BUSI)

Quality of the + neighborhood business environment, graded by sz.fang.com, ranging from 1 (lowest) to 100 (highest)

Wen et al. (2014, 2019)

Zheng and Kahn (2008, 2013), Wen et al. (2019) −

Chau and Ng (1998), Chau and Chin (2003), Yang et al. (2019)

(continued)

16

2 Place-Varying Impact of Metro Accessibility on Property Prices

Table 2.2 (continued) Classification

Explanatory variables

Variable (code)

Description

Expected sign

Reference

Shopping environment (SHOP)

Quality of the neighborhood shopping environment, graded by sz.fang.com, ranging from 1 (lowest) to 100 (highest)

+

Chau and Ng (1998), Chau and Chin (2003), Yang et al. (2019)

Recreation facilities (REC)

Quality of + neighborhood recreation facilities, graded by sz.fang.com, ranging from 1 (lowest) to 100 (highest)

Chau and Chin (2003), Zheng and Kahn (2008)

Distance to URT station (DMETRO)

Network distance to the nearest station on Line 5 (m)

Distance to URT station within 0–400 m (DMETRO_1)

Dummy variable, + equals one for a property within 400 m of URT stations and zero otherwise

Distance to URT station within 400–800 m (DMETRO_2)

Dummy variable, equals one for a property within 400–800 m of URT stations and zero otherwise

+

Distance to URT station within 800–1200 m (DMETRO_3)

Dummy variable, equals one for a property within 800–1200 m of URT stations and zero otherwise

+

Distance to URT station within 1200–1600 m (DMETRO_4)

Dummy variable, equals one for a property within 1200–1600 m of URT stations and zero otherwise

+



Chau and Ng (1998), Chau and Chin (2003), Zheng and Kahn (2008, 2013)

(continued)

2.4 Property Price Data

17

Table 2.2 (continued) Classification

Variable (code)

Description

Distance to URT station within 1600–2000 m (DMETRO_5)

Dummy variable, equals one for a property within 1600–2000 m of URT stations and zero otherwise

Proximity to transfer station (TRANS)

Dummy variable, equals one for a property within 2000 m of transfer stations and zero otherwise

Expected sign

Reference

+

Dai et al. (2016)

Table 2.2 presents the definitions, quantifications, and expected signs of variables. Typical characteristics used for hedonic analysis can be categorized into structural, locational, and neighborhood characteristics. Therefore, a total of ten control variables are selected, including five structural, one locational, and four neighborhood characteristics. Moreover, this study aims to test the sub-market effect of URT stations on property values. Therefore, the sample is further divided into two groups, namely the sub-sample in the urban area (“urban area sub-sample” hereafter, N = 325) and that in the suburban (“suburban area sub-sample” hereafter, N = 397). The descriptive statistics of the variables of the full sample and the two sub-samples are shown in Table 2.3. Note that distance variables are measured within ESRI ArcGIS software (v 10.0).

2.5 Results A pairwise correlation matrix was first calculated to showcase the correlation between the independent variables. Results (not shown here) show that collinearity is not an issue for the data.

2.5.1 Basic Models: Does URT Accessibility Offer Premiums? Table 2.4 shows the results of two basic models for the overall sample. Note that the only difference between Models 1 and 2 is the type of explanatory variables: Model 1 uses a continuous variable, while Model 2 uses dummy variables. Regarding the control variables, the coefficients associated with variables of AGE, FEE, GREEN, EDU, BUSI, and SHOP are significant at the 10% level. This implies

18

2 Place-Varying Impact of Metro Accessibility on Property Prices

Table 2.3 Descriptive statistics of the dataset Overall sample (N = 722)

Urban area sub-sample (N = 325)

Suburban area sub-sample (N = 397)

Mean

Std. Dev.

Mean

Std. Dev.

Mean

Std. Dev.

P

50,171

9830

53,750

9465

47,730

9562

AGE

17.95

7.39

20.28

7.53

15.74

6.55

FEE

2.16

1.12

2.39

1.17

1.97

1.04

FAR

3.92

2.63

4.66

3.02

3.28

2.02

GREEN

0.32

0.13

0.3

0.13

0.33

0.12

PARK

0.82

0.93

0.6

0.53

0.97

1.1

DCBD

14,830

4674

11,677

2629

17,813

4205

EDU

2.67

1.91

3.08

1.96

2.27

1.78

BUSI

52.07

28.59

56.61

33.97

47.77

21.55

SHOP

78.04

24.54

66.48

28.81

88.97

11.96

REC

84.38

23.58

77.83

29.98

90.57

12.44

DMETRO

1031.3

355.4

946.7

357.6

1111.5

342.9

DMETRO_1

0.09

0.29

0.06

0.23

0.13

0.33

DMETRO_2

0.26

0.44

0.21

0.41

0.31

0.46

DMETRO_3

0.3

0.46

0.34

0.48

0.26

0.44

DMETRO_4

0.21

0.41

0.22

0.41

0.2

0.4

DMETRO_5

0.14

0.35

0.17

0.38

0.1

0.31

TRANS

0.22

0.41

0.33

0.47

0.12

0.32

Variable

that these variables significantly impact the prices of properties located within 2 km of URT. However, DCBD, FAR, PARK, and REC have negligible effects on property values. Furthermore, the signs of all the variables meet our expectations. The interpretation of the URT accessibility variables is of great interest here. In Model 1, lnDMETRO and lnDMETRO2 are significant at the 5% level, and the coefficients associated with lnDMETRO and lnDMETRO2 are positive and negative, respectively. This suggests a non-linear effect of distance to URT station on property prices. This result implies that the value uplift effects of URT disappear quickly as the distance to the URT station increases. Moreover, in Model 2, the four dummy variables are significant at the 10% level, indicating that URT accessibility has significant positive impacts on surrounding residential property values. These coefficients generally decrease with advancing distance to URT stations, and the only exception is that the coefficient of DMETRO_2 is larger than that of DMETRO_1. This observation suggests that URP price premiums quintessentially decrease with advancing distance and that the largest influence occurs in the area within 400–800 m from URT stations. Regarding goodness of fit, the adjusted R2 of Models 1 (0.501) and 2 (0.561) are satisfactory. This observation suggests that the independent variables

2.5 Results

19

Table 2.4 Regression results of basic models Variable

Model 1 Coefficient

Model 2 t-statistic

Coefficient

t-statistic

AGE

−0.004*

−1.90

−0.004**

−2.08

FEE

0.067***

5.21

0.063***

4.89

FAR

−0.007*

−1.68

−0.007

−1.59

GREEN

0.184**

2.00

0.194**

2.14

PARK

0.003

0.28

0.004

0.11

lnDCBD

−0.055

−1.36

−0.055

−1.38 2.07

EDU





0.087**

lnBUSI

0.194***

8.96

0.194***

9.01

lnSHO

0.103*

1.91

0.113*

1.82

lnREC

0.025

1.02

0.021

0.91

lnDMETRO

0.814**

2.08





lnDMETRO2

−0.066**

−2.05



– 1.99

DMETRO_1





0.068**

DMETRO_2





0.079**

2.03

DMETRO_3





0.051*

1.85 1.79 15.04

DMETRO_4





0.043*

Constant

9.492***

9.70

10.406***

Performance statistic Adjusted R2

0.501

0.561

Note *** Significant at the 1% level. ** Significant at the 5% level. * Significant at the 10% level

of these two models can explain roughly 50.1% and 56.1%, respectively, of variation in the dependent variable. The goodness-of-fit values are close to those of other housing-project-level hedonic pricing models (Zheng & Kahn, 2008, 2013).

2.5.2 Enhanced Model: Do Transfer Stations Provide Larger Accessibility Benefits Than Regular Stations? We focus on the impacts of transfer and regular stations on residential property prices by introducing an additional variable of TRANS into Model 1. Table 2.5 presents the regression results. The signs of all variables are reasonable, and 10 of the 13 independent variables are significant at the 10% level. Moreover, the goodness-of-fit has increased from 0.501 to 0.578 due to the addition of the variable TRANS. As expected, transfer stations have a larger value uplift effect on residential properties than regular stations. This finding is in accordance with Dai et al. (2016).

20 Table 2.5 Regression results of the enhanced model

2 Place-Varying Impact of Metro Accessibility on Property Prices Variable

Coefficient

t-statistic

AGE

−0.004*

−1.91

FEE

0.066***

5.06

FAR

−0.008*

−1.71

GREEN

0.182**

1.97

PARK

0.003

0.27

lnDCBD

−0.065

−1.60

EDU

0.090**

2.02

lnBUSI

0.189***

8.62

lnSHOP

0.097*

1.92

lnREC

0.023

0.96

lnDMETRO

0.832**

2.03

lnDMETRO2

−0.067**

−2.07

TRANS

0.041*

1.65

Constant

9.083***

9.00

Performance statistic Adjusted R2

0.578

Note *** *

Significant at the 1% level. ** Significant at the 5% level. Significant at the 10% level

2.5.3 Market Segmentation Analyses: Are URT Accessibility Benefits More Perceptible in the Suburban Area Than in the Urban Area? Compared with the urban area, URT accessibility benefits may be more perceptible in the suburban area where the public transport service is relatively limited. To test this hypothesis, we divided the sample into two groups according to the location (within or outside the original SEZ), namely the urban (guannei) and suburban areas (guanwai). A Chow test (Chow, 1960) is conducted to verify the existence of market segmentation, and the corresponding results are presented in Table 2.6. Chow statistics for Models 1 and 2 are 1.95 and 1.83, respectively. This result rejects the null hypothesis at the 95% confidence level and supports the validity of market segmentation. As such, we have sufficient evidence to conclude that different behaviors can Table 2.6 Chow test results

Model

Chow statistic

Result

Market segmentation

Model 1

1.83

Reject the null hypothesis

Yes

Model 2

1.95

Reject the null hypothesis

Yes

2.5 Results

21

be observed in the two sub-markets and that the two sub-samples in the urban and suburban areas of Shenzhen cannot be pooled together. The effects of URT accessibility on property prices in the urban and suburban property markets are evaluated using separate hedonic pricing models, and both continuous and dummy measures of URT accessibility are used again. The results are shown in Table 2.7. The signs of all the variables are as expected in both the urban and suburban markets. Specifically, the independent variables can explain 53–63% of variations in property prices. On the one hand, the coefficients of continuous distance variables (i.e., lnDMETRO and lnDMETRO2 ) for the two sub-samples are significant at the 5% level, indicating that an inverted-U and non-linear relationship between URT and property prices. On the other hand, dummy variables that describe URT accessibility (i.e., DMETRO_1, DMETRO_2, DMETRO_3, and DMETRO_4) for the suburban market have larger coefficients and t-statistics than those for the urban market. This result indicates that URT accessibility benefits are more perceptible in the suburban area. Moreover, the variable DMETRO_4 is significant at the 10% level in the suburban market but not significant in the urban counterpart, meaning that the influencing range of URT is larger in the suburban area (1200 m for the urban area and 1600 m for the suburban area). This outcome can be explained by the fact that residents in the suburban area have fewer travel mode choices and thus are more willing to pay for URT accessibility. Moreover, the difference between transfer and regular stations is larger in the suburban area than in the urban area, which is in line with Kim and Zhang (2005). All in all, our outcomes support that URT accessibility benefits are more notable in the suburban area than in the urban area.

2.6 Discussion In China, URT development normally neither receives funding from the central and provincial governments (Yang et al., 2016) nor captures the property price increment attributed to URT provision. It mainly relies on land auction prices and bank debts now. Implementing value capture schemes to avoid free-ride or let residents (the owners of properties in the URT catchment area) squeeze some values gained from URT provision is essential. Property tax is a common value capture tool used in many countries worldwide. However, at present, all cities in China other than Shanghai and Chongqing do not levy such a tax. Moreover, local governments in Mainland China do not have the legislative power to tax and thus cannot learn from and adopt the Shanghai/Chongqing model. We feel that property tax can be tested in more cities in China due to differences in the institutions of China and countries with property taxes. In other words, China faces institutional barriers in using the property tax tool. Directly transferring property tax models from abroad to China must be improper. Therefore, we think that China-specific schemes should be hotly debated and evaluated in more cities in the upcoming years. At present, very few cities (e.g., Shenzhen and Wuhan) have put efforts into this issue and used innovative but informal value capture strategies, such as the “R + P” program, two-step bid-ding, and

1.71

1.15

2.96

1.98

1.82

0.29

2.11

−2.09



1.95

2.41

0.262*

0.008

0.012

0.080***

0.190**

0.072*

0.033

0.739**

−0.060**









0.020*

5.986**

FAR

GREEN

PARK

lnDCBD

EDU

lnBUSI

lnSHOP

lnREC

lnDMETRO

lnDMETRO2

DMETRO_1

DMETRO_2

DMETRO_3

DMETRO_4

TRANS

Constant

Note

***

0.592

Significant at the 1% level.

Adjusted R2

**

−2.01

−0.012*

FEE

Performance statistic

2.21

0.055**

0.629

5.537***

0.049**









−0.087***

1.100***

0.028

0.087*

0.208***

0.097**

*

3.18

2.36









−3.00

2.99

0.98

1.89

3.41

2.02

0.81 −1.65

0.009

2.02

−0.12

2.87

−2.49

t-statistic

−0.141

0.242**

−0.001

0.049***

−0.008**

2.05 1.83

0.073** 0.048*

0.530

6.364***



2.94



1.62

1.92

0.065*

0.039





0.78

1.79

5.98

2.69

−1.11

0.19

1.79

−2.04

2.22

−1.82

t-statistic





0.020

0.067*

0.201***

0.079***

−0.060

0.006

0.268*

−0.012**

0.054**

−0.005*

Coefficient

Urban area sub-sample

Significant at the 5% level. Significant at the 10% level







0.25

−1.87

Coefficient

t-statistic

Coefficient

−0.005*

Suburban area sub-sample

Urban area sub-sample

AGE

Variable

Table 2.7 Regression results of the urban and suburban area sub-samples

0.630

8.511***

6.84



1.85 –

1.94 0.045*

2.66

2.02





0.85

1.91

3.19

2.03

−0.96

0.51

1.94

−0.16

2.95

−2.36

t-statistic

0.066*

0.088***

0.071**





0.018

0.111*

0.181***

0.092**

−0.106

0.006

0.231*

−0.001

0.050***

−0.008**

Coefficient

Suburban area sub-sample

22 2 Place-Varying Impact of Metro Accessibility on Property Prices

2.6 Discussion

23

land reserve (Wang et al., 2019). Furthermore, most cities can test the suitability of these China-specific strategies and explore new approaches, such as predetermined land reserve (Sun et al., 2017), for gaining revenues to enhance public services and eventually the life quality of citizens. Considering the different impacts of URT on property values in the urban and suburban areas, we suggest that local governments adopt diversified value capture schemes (specific for each area) instead of a universal scheme that applies to the whole city and also promote URT-oriented-development in the suburban area to effectively reap the benefits offered by URT. Furthermore, transfer stations offer larger price premiums than regular stations. On this basis, we suggest that decisionmakers introduce additional activities, services, and opportunities in the adjacent areas of transfer stations.

2.7 Conclusions URT provides the opportunity to mitigate numerous urban problems (e.g., traffic congestion and degradation of the environment) and thus has attracted substantial scholarly attention in recent years. URT has been extensively promoted in today’s China, which takes greater responsibility both as an economic powerhouse and a sustainability advocator. Typically, as a stimulus or catalyst for driving economic development and encouraging urbanization, URT enhances the transportation mobility of its adjacent areas and raises the prices of nearby properties. The impacts of URT accessibility on property values have been well documented in a voluminous body of literature, but only a few studies have considered differences in the price effects of URT transfer and regular stations and differences between the urban and suburban areas. Based on 722 residential complexes within the distance of 2 km from Line 5 in Shenzhen, China, a set of hedonic pricing models is developed to investigate the correlation between URT accessibility and property prices. The main findings of this study are as follows. (1) URT is positively related to nearby property prices, and its price effects extend to 1.6 km from stations. (2) The positive impact diminishes quickly as the distance to the URT station increases. (3) Transfer stations provide larger accessibility benefits than regular ones. (4) Benefits stemming from URT accessibility are more notable in the suburban area than in the urban area, and the influencing geographical range is larger in the suburban area. The first two findings are consistent with the majority of relevant studies. By contrast, the last two findings have seldom been identified in the existing literature and thus serve as valuable contributions to the knowledge about URT systems and property values. Several limitations merit future research. First, the influencing geographical range of URT derived from this study seems to be larger than that reported in the existing literature (c.f., Diao et al., 2017; Li, 2018). We hypothesize that this difference is due to the ubiquitous use of bike-sharing in the city (Lin et al., 2019). More empirical

24

2 Place-Varying Impact of Metro Accessibility on Property Prices

studies, therefore, can be conducted to test this new hypothesis in bike-sharingdominant cities, such as Mainland Chinese cities and Singapore. Second, this study only focuses on residential properties; on the contrary, value capture schemes may be more feasible for high-end office and retail properties (Cervero & Susantono, 1999). Thus, we recommend that future studies be devoted to discussing the relationship between URT accessibility and office and/or retail property prices. Last, we carefully considered various issues in selecting the study area and finally selected the Line 5 adjacent area for testing our two hypotheses on place-varying price effects of URT. The selection of the URT adjacent area used for the statistical analysis also follows the existing literature. This study provides interesting modeling outcomes and offers evidence supporting the place-varying effects of URT. Moreover, this study draws practical implications. The generality and transferability of its findings, however, are unknown. Notably, our empirical study is not alone in this regard. In other words, most, if not all, empirical researchers are often confused about whether conclusions drawn from one area can be applied to other areas. As such, we suggest that conducting more sophisticated empirical studies in other areas is indispensable for us to reach safer and more persuasive conclusions.

References Bartholomew, K., & Ewing, R. (2011). Hedonic price effects of pedestrian-and transit-oriented development. Journal of Planning Literature, 26(1), 18–34. Bowes, D. R., & Ihlanfeldt, K. R. (2001). Identifying the impacts of rail transit stations on residential property values. Journal of Urban Economics, 50(1), 1–25. Cervero, R., & Susantono, B. (1999). Rent capitalization and transportation infrastructure development in Jakarta. Review of Urban and Regional Development Studies, 11(1), 11–23. Chau, K. W., & Chin, T. L. (2003). A critical review of literature on the hedonic price model. International Journal for Housing Science and Its Applications, 27(2), 145–165. Chau, K. W., & Ng, F. (1998). The effects of improvement in public transportation capacity on residential price gradient in Hong Kong. Journal of Property Valuation and Investment, 16, 397–410. Chow, G. C. (1960). Tests of equality between sets of coefficients in two linear regressions. Econometrica, 28(3), 591–605. Dai, X., Bai, X., & Xu, M. (2016). The influence of Beijing rail transfer stations on surrounding housing prices. Habitat International, 55, 79–88. Diao, M., Leonard, D., & Sing, T. F. (2017). Spatial-difference-in-differences models for impact of new mass rapid transit line on private housing values. Regional Science and Urban Economics, 67, 64–77. Hill, R. J. (2013). Hedonic price indexes for residential housing: A survey, evaluation and taxonomy. Journal of Economic Surveys, 27, 879–914. Kim, J., & Zhang, M. (2005). Determining transit’s impact on Seoul commercial land values: An application of spatial econometrics. International Real Estate Review, 8(1), 1–26. Ko, K., & Cao, X. J. (2013). The impact of Hiawatha Light Rail on commercial and industrial property values in Minneapolis. Journal of Public Transportation, 16(1), 47–66. Krause, A. L., & Bitter, C. (2012). Spatial econometrics, land values and sustainability: Trends in real estate valuation research. Cities, 29, S19–S25.

References

25

Lancaster, K. J. (1966). A new approach to consumer theory. Journal of Political Economy, 74, 132–157. Li, Z. (2018). The impact of metro accessibility on residential property values: An empirical analysis. Research in Transportation Economics, 70, 52–56. Lin, B., & Du, Z. (2017). Can urban rail transit curb automobile energy consumption? Energy Policy, 105, 120–127. Lin, D., Zhang, Y., Zhu, R., & Meng, L. (2019). The analysis of catchment areas of metro stations using trajectory data generated by dockless shared bikes. Sustainable Cities and Society, 49, 101598. Malpezzi, S. (2003). Hedonic pricing models: A selective and applied review. Housing Economics and Public Policy, 67–89. Rosen, S. (1974). Hedonic prices and implicit markets: Product differentiation in pure competition. The Journal of Political Economy, 82(1), 34–55. Sharma, R., & Newman, P. (2018). Can land value capture make PPP’s competitive in fares? A Mumbai case study. Transport Policy, 64, 123–131. Sirmans, S., Macpherson, D., & Zietz, E. (2005). The composition of hedonic pricing models. Journal of Real Estate Literature, 13(1), 1–44. Sun, J., Chen, T., Cheng, Z., Wang, C. C., & Ning, X. (2017). A financing mode of Urban Rail transit based on land value capture: A case study in Wuhan City. Transport Policy, 57, 59–67. Tsai, H., Huang, W. J., & Li, Y. (2016). The impact of tourism resources on tourism real estate value. Asia Pacific Journal of Tourism Research, 21(10), 1114–1125. Wang, J., Samsura, D. A. A., & van der Krabben, E. (2019). Institutional barriers to financing transitoriented development in China: Analyzing informal land value capture strategies. Transport Policy, 82, 1–10. Wen, H., Zhang, Y., & Zhang, L. (2014). Do educational facilities affect housing price? An empirical study in Hangzhou, China. Habitat International, 42, 155–163. Wen, H., Xiao, Y., & Hui, E. C. (2019). Quantile effect of educational facilities on housing price: Do homebuyers of higher-priced housing pay more for educational resources? Cities, 90, 100–112. Yang, J., Chen, J., Le, X., & Zhang, Q. (2016). Density-oriented versus development-oriented transit investment: Decoding metro station location selection in Shenzhen. Transport Policy, 51, 93–102. Yang, L., Wang, B., Zhou, J., & Wang, X. (2018). Walking accessibility and property prices. Transportation Research Part D: Transport and Environment, 62, 551–562. Yang, L., Zhou, J., Shyr, O. F., & Huo, D. (2019). Does bus accessibility affect property prices? Cities, 84, 56–65. Yang, L., Chau, K. W., Szeto, W. Y., Cui, X., & Wang, X. (2020a). Accessibility to transit, by transit, and property prices: Spatially varying relationships. Transportation Research Part D: Transport and Environment, 85, 102387. Yang, J., Zhu, L., Duan, Y., Zhou, J., & Ma, H. (2020b). Developing metro-based accessibility: Three aspects of China’s Rail+Property practice. Transportation Research Part D: Transport and Environment, 81, 102288. Zheng, S., & Kahn, M. E. (2008). Land and residential property markets in a booming economy: New evidence from Beijing. Journal of Urban Economics, 63(2), 743–757. Zheng, S., & Kahn, M. E. (2013). Does government investment in local public goods spur gentrification? Evidence from Beijing. Real Estate Economics, 41(1), 1–28.

Chapter 3

Time-Varying Impact of Metro Accessibility on Property Prices

3.1 Introduction Driven by forces such as economic upswings and long holiday durations, tourism demand worldwide has dramatically increased in recent years. For example, global international tourist volume (visitor arrivals) in 2018 saw a 5.4% increase, compared with 2017; and the growth of global international tourism receipts from 2017 to 2018 was 4.4%, while world GDP growth during the same period was only 3.0%. Shopping is one of the most important tourism activities in tourist destinations, especially shopping destinations (e.g., Paris, Hong Kong, Singapore, Seoul, and Dubai) (Choi et al., 2016; Xu & McGehee, 2012). On the one hand, shopping fulfills the utilitarian needs of tourists (purchasing miscellaneous necessities for daily needs and duty-free goods). On the other hand, shopping provides a precious opportunity for tourists to expose to the host culture and offers tourists a fruitful hedonic, recreational, and touristic experience (fleeing from mundane routine and buying souvenirs and artworks as reminders of the travel experience) (Meng & Xu, 2012). As such, in many cases, shopping is a major reason behind travel (Meng & Xu, 2012). More importantly, it is especially essential in the current era of materialism and consumption. Understanding the linkages between tourism demand and the retail property (or retail shop) market in a shopping destination is of paramount importance. The existing literature, however, has inadequately delved into the linkages. Three exceptions are the work of Li et al. (2018), Yang et al. (2020), and Jayantha and Yung (2018). Li et al. (2018) utilized the street-level retail property transaction data to examine the impacts of cross-border tourist shoppers on the retail property market of Hong Kong and found that the policy Multiple-entry Permit leads to the increase in the prices of retail properties, especially those of new and large-sized properties. Yang et al. (2020) adopted standard and error-correction-model-based Granger causality tests to examine the relationships between tourism development and retail property prices between 2002Q1 and 2014Q4 in Hong Kong. The authors concluded that tourism development Granger causes an increase in retail property prices in the popular © The Author(s), under exclusive license to Springer Nature Singapore Pte Ltd. 2021 L. Yang, Property Price Impacts of Environment-Friendly Transport Accessibility in Chinese Cities, https://doi.org/10.1007/978-981-16-8833-1_3

27

28

3 Time-Varying Impact of Metro …

tourism shopping area, but not in the unpopular. Jayantha and Yung (2018) estimated semi-log hedonic pricing models for the rentals of ground-floor retail properties in the old area of Wanchai, Hong Kong, and observed that revitalized historical projects are positively associated with the rentals of nearby retail properties. Moreover, they stated that gross floor area and transit accessibility contribute to explaining the rentals of ground-floor retail properties in the old area of Wanchai, Hong Kong, while age and property management body are too weak to determine the rentals. How does tourism demand (specifically, the substantial increase in tourist shoppers permitted by the IVS) affect the retail property market? Does the increase in tourist shoppers alter the implicit prices of certain retail property attributes (or characteristics) (e.g., transit accessibility and accommodation facility accessibility)? These questions are insufficiently answered by existing studies and thus are what this study attempts to probe into (explained in Sect. 3.3). To address the above issues, this study examines the association between tourism demand (specifically, the increase in tourist shoppers) and retail property prices in Hong Kong under the hedonic framework. This is viable for the following two reasons: (1) The IVS can be used as a quasi-natural experiment to investigate the effect of the increasing volume of tourist shoppers on the pricing of retail properties; and (2) more importantly, Hong Kong has an active traded and transparent market for street-level retail properties. In addition, given that the presence of spatial autocorrelation is a major problem in property price modeling, this study estimates the traditional hedonic pricing model and spatial econometric models to tackle the spatial autocorrelation problem and tests our hypotheses regarding tourism demand and street-level retail property prices (explained in Sect. 3.3). The contributions of this study include: (1) investigating how tourism demand affects the pricing of street-level retail properties in a shopping destination; (2) inferring the behavior of tourist shoppers from transaction prices of street-level retail properties as an alternative to the questionnaire survey or interview; (3) using a policy change, that is, the implementation of the IVS, as a quasi-natural experiment to test a number of hypotheses concerning the behavior of tourist shoppers; and (4) contributing to the hot debate on the interaction between tourism development and the retail property market. The remainder of the paper is organized as follows. Section 3.2 offers an overview of the study area, Hong Kong. Section 3.3 develops a set of hypotheses. Section 3.4 describes the retail property price data. Section 3.5 introduces the spatial Durbin model. Section 3.6 reports and discusses the empirical results. Section 3.7 concludes the paper, discusses potential practical implications, and identifies research limitations.

3.2 Overview of Hong Kong Hong Kong has an established worldwide reputation for being a “shopping paradise” and is a typical shopping destination, especially for tourists from Mainland China. According to the Hong Kong Tourism Board, shopping constituted the majority

3.2 Overview of Hong Kong

29

of travel expenditure: 86.7% for same-day visitors (or day-trippers) and 51% for overnight visitors in 2018. This demonstrates that shopping appeals to the majority of inbound tourists. A trip to Hong Kong is deemed incomplete without shopping activities (Choi et al., 2016). Moreover, Hong Kong is a quintessential compact, high-density, and transit-oriented city. The Mass Transit Railway (MTR) is the most extensively used travel mode, which takes the largest market share (over 30% of trips). Furthermore, it is more frequently used by tourists than local residents. In Hong Kong, the implementation of the policy Individual Visit Scheme (IVS) has introduced numerous tourist shoppers from Mainland China. The IVS was initially launched on July 28, 2003, under the Closer Economic Partnership Arrangement between Mainland China and Hong Kong. The IVS is a tourism liberalization and tourism–promotion strategy that allows eligible residents with permanent household registration in specific Mainland Chinese cities to visit Hong Kong individually. Before the implementation of the IVS, Mainland Chinese visitors must apply for business visas or group-based tours for a Hong Kong visit.

3.3 Hypothesis Development According to Lindquist (1974), shopping satisfaction is affected by attributes of the retail space, such as convenience, physical environment, and atmosphere. Some attributes of the retail space can be reflected by and measured with corresponding variables. This provides the theoretical foundation for linking the increase in tourist shopper volume with retail property transaction prices. Thus, we aim to investigate tourist shoppers’ behavior from the perspective of the pricing of the shopping space instead of relying on the commonly-used research approach, namely questionnaire surveys. Identifying utility-bearing and price-influencing attributes of retail properties in the tourist precinct of a shopping destination helps reveal the preferences of tourist shoppers (primary targets of such retail properties). Consumer behavior is essentially rational: consumers often carefully consider or evaluate the costs and benefits of each possible choice before making the final decision. Economic factors are highly important in choice evaluation. As such, in theory, tourist shoppers highly care about economic factors, such as travel cost (including time and monetary costs), and may highly value transport accessibility (e.g., accessibility to transit) (Xu & Yang, 2019; Yang et al., 2019) when selecting where to shop. For tourist shoppers, the time of stay in a tourist destination is relatively short. Therefore, tourist shoppers have a much higher value-of-time than local shoppers. Tourist shoppers tend to minimize the time of accessing shopping attractions and consequently maximize their valuable shopping time (which derives utility). Therefore, tourist shoppers are more willing to shop in accessible locations or locations with a high level of transport accessibility than local shoppers (Kang, 2016). In Hong Kong, one easily observable and quantifiable indicator of geographical accessibility is the accessibility to the MTR. Generally, accessibility to the MTR

30

3 Time-Varying Impact of Metro …

has positive effects on retail property prices, given that MTR travel often saves consumers’ travel time and monetary cost of reaching shopping opportunities. The demand for locations with high metro accessibility should increase if the tourist shopper volume increases. In other words, after the implementation of the IVS (which induces a huge number of tourists), the benefits (or price effects) of accessibility to the MTR should, in theory, increase. Based on the above reasoning, we formulate the first hypothesis (H1): H1: The implicit price of accessibility to the MTR would increase after the implementation of the IVS.

Accommodation facilities in a tourist destination (e.g., hotels or guesthouses) offer paid lodging to people on a short-term basis. They are always a big issue, especially for overnight tourist shoppers. Tourist shoppers have a higher cost of stay and a tighter schedule than local shoppers. Thus, tourist shoppers care more about accommodation issues. A retail property located near accommodation facilities reduces travel costs, increases shopping time, and eases shopping activities. Retail properties with good accessibility to accommodation facilities may have a positive effect on tourist shoppers’ shopping experience. Similar to the reasoning of H1, after the implementation of the IVS, the benefits of accessibility to accommodation facilities may increase. However, in Hong Kong, the majority (approximately 60%) of tourists from Mainland China (beneficiary of the IVS) are same-day in-town visitors. They do not stay overnight and thus are too transient to utilize accommodation facilities. Moreover, the accommodation cost in tourist precincts (or popular tourism shopping area) of a shopping destination, such as Causeway Bay in Hong Kong (the study area of this study), is, in general, extraordinarily high. As such, only a small proportion of tourist shoppers stay within the tourist precincts. Thus, the implicit price of accessibility to accommodation facilities may not be significantly altered by the implementation of the IVS. Accordingly, we propose two contrasting or competing hypotheses (H2A and H2B): H2A: The implicit price of accessibility to accommodation facilities would increase after the implementation of the IVS. H2B: The implicit price of accessibility to accommodation facilities would not be altered by the increase after the implementation of the IVS.

Besides the above economic concerns on transportation and accommodation, psychological factors, such as the hometown shopping experience, can affect tourists’ shopping behavior. People (including tourist shoppers) often make unconscious decisions based on experience and sometimes even have done that without realizing the involvement of memories in the decision-making process. The attitudes of tourist shoppers (mainly those from Mainland China) toward the shopping environment in tourist destinations can be affected by their hometown experience. Mainland China has a short history of urbanization. In Mainland China, reputable retailers are mainly located in newly built shops. By contrast, old shops

3.3 Hypothesis Development

31

usually attract non-reputable retailers, so the chance of getting counterfeited items is high in such shops. Therefore, tourist shoppers from Mainland China can easily associate shops located in old buildings with low-quality retailers based on their local experience and psychological implications in home areas. In other words, tourist shoppers will take such shops as a signal of low quality and service. By contrast, local shoppers have much more information to make decisions than tourist shoppers and rarely do so. Old shops reduce the shopping intention of tourist shoppers, which indirectly affects the sales price of retail properties. Thus, we propose the third hypothesis (H3): H3: Age would have a larger negative price effect after the implementation of the IVS.

3.4 Snapshot of the Spatial Durbin Model As noted in Chap. 2, the hedonic pricing model is a popular method for empirically assessing and identifying the determinants of the prices of heterogeneous products such as properties. In spite of the highly recognizable usefulness of hedonic price models, they still have a few problems, such as misspecification of functional form, omitted (or missing) variable bias, and ignorance of spatial dependence (or spatial autocorrelation), which could lead to biased and/or inefficient estimates. First, for many problems, traditional economic theories fail to suggest a “correct” or “best” model specification. Thus, the hedonic price model potentially takes a host of functional forms, such as linear, semi-log (or log-linear), double-log, translog, and semi-log quadratic. Thus, choosing functional forms should be treated with caution. Normally, non-linear functional forms (e.g., double-log and Box-Cox transformation) are preferred. Second, hedonic price models directly control for property characteristics by inserting them into the regression models. In theory, all contributory variables should be included. However, this cannot be achieved in practice. Therefore, hedonic price models cannot tell the whole story. As such, it is a crucial decision to choose independent variables that can truly affect the value of properties. Third, the price/rental of a property affects and is also affected by the prices/rentals of other properties situated relatively near. This phenomenon is often referred to as spatial dependence, for which alternative explanations include spatial externality, spatial interaction, and external force (Li et al., 2018). Inevitably, the use of spatial data often fails to meet the underlying Gauss-Markov assumption of OLS regression. As introduced before, the hedonic pricing model can be expressed as follows. Y = αln + Xβ + ε, where Y denotes an n × 1 vector of property price; n is the number of observations; ln is an n × 1 vector of ones associated with the constant α; X is an n × k matrix of hedonic characteristics of the property (e.g., size, age, and distance to the nearest transit station); β is a k × 1 vector of coefficients; and ε is an n × 1 vector of random

32

3 Time-Varying Impact of Metro …

error terms that follow a normal distribution. α and β are parameters to be jointly estimated by the ordinary least squares (OLS) method. A host of spatial econometric models, such as the spatial lag model (SLM), the spatial error model (SEM), and the spatial Durbin model (SDM), can address the spatial autocorrelation issue, and they have been widely used to explain the relationship between property prices and property characteristics. The SLM and the SEM are two basic spatial econometric models and focus on the endogenous interaction relationship (or spatial interaction in the dependent variable) and the correlated relationship (or spatial interaction in the error term), respectively. They, however, cannot address the exogenous interaction relationship (or spatial interaction in independent variables) (Manski, 1993). The SDM jointly considers the spatial dependence of the dependent variable and that of independent variables and allows for prices and hedonic characteristics of nearby properties to shape the price of a specific property. It can be specified as follows: Y = ρW Y + αln + Xβ + W X θ + ε, where W is an n × n spatial weight matrix, exogenously defined by either contiguity or distance; W Y is the spatially lagged property price; ρ is the spatial autoregressive parameter, representing the price effect of W Y ; W X is the spatially lagged hedonic characteristics; θ is a k × 1 vector of coefficients, reflecting the price effects of W X ; and other variables are as defined before. α, β, ρ, and θ are parameters to be jointly determined by the maximum likelihood estimation method. ρW Y and W X θ capture endogenous and exogenous interaction relationships, respectively.

3.5 Property Price Data To avoid the influence of numerous hard-to-control and even unobservable confounding attributes, especially locational and environmental attributes, on retail property prices, focusing on a small geographical area is better than the entire city and a large area (e.g., Hong Kong Island and Kowloon). Causeway Bay is among the world’s most expensive retail property locations. Moreover, according to the Transport Department of Hong Kong, Causeway Bay attracts many trips of inbound tourists. Therefore, Causeway Bay, the area widely recognized as the tourist precinct of Hong Kong that has the long-known worldwide reputation “shopping paradise,” is chosen as the study area. The transaction price rather than the rental (which is used in a voluminous body of previous literature) of retail properties is used as the dependent variable in this study. The reasons are as follows: (1) the rental of retail properties is determined not only by property attributes but also by other factors, including leasing terms and type of tenant (Benjamin et al., 1990); and (2) the rental is often composed of minimum (or base) and average (or percentage) rentals (Eppli & Benjamin, 1994). The feature

3.5 Property Price Data

33

Fig. 3.1 Geographical distribution of the retail property samples

makes property rental modeling extremely difficult. Understandably, obtaining the details of each rental contract is impractical. Moreover, there is no available lease-bylease retail rental data in Hong Kong. Thus, the transaction price is a more reliable, consistent, and available measure than the rental, especially in the context of Hong Kong. The transaction records of ground-floor retail properties in Causeway Bay within the period of 1993–2011 serve as our database. The transaction price data are purchased from the Economic Property Research Center (EPRC). The total number of retail property transaction observations during the time frame is 3806. After excluding observations with missing information, 580 transaction observations are left. The samples used for the subsequent analysis are mapped in Fig. 3.1. It is widely recognized that property price is widely influenced by a multitude of variables. Independent variables used in the existing hedonic literature are often categorized into three categories: structural (e.g., size and age), locational (e.g., access to the downtown), and neighborhood (e.g., landscape view and crime). In this study, the selection of variables is mainly informed by previous studies while taking data availability into consideration. Mainland China is the largest sub-market for the Hong Kong tourism industry, and IVS tourist shoppers hold a great consumption capacity. Therefore, three IVS-relevant variables are incorporated into the retail property price modeling framework to test the three sets of hypotheses. The detailed description of variables is listed in Table 3.1. The last three interaction variables are used for hypothesis testing. Table 3.2 shows the summary statistics of the main variables. The area of the smallest retail property (which may be used as a fruit or take-away beverage shop) is only 5 m2 . In addition to the property price data from the EPRC, other data were collected from the Hong Kong Tourism Board, the Census and Statistics, and the

34

3 Time-Varying Impact of Metro …

Table 3.1 Variables and description Variable

Description

Expected sign Remark

LnP

Logarithm of the transaction price (in natural logarithm form) (HK$)

NA

Dependent Variable

AGE

(year)

?

Control

SIZE

Size or gross floor area (m2 )

+

Control

SIZE2

Square term of SIZE

?

Control

FRON

Length of frontage facing the street (m)

+

Control

LnMTR

Logarithm of distance to the nearest MTR station exit (m) (in natural logarithm form)



Control

LnMALL

Logarithm of distance to the nearest shopping mall (m) (in natural logarithm form)



Control

CORN

Dummy variable, equals one if the property is located in the street corner and zero otherwise

+

Control

ACM

Number of hotels and guesthouses within the 250 m radius

+

Control

ACM2

Square term of ACM



Control

UCU

Dummy variable, equals one if the + property’s upper story is for commercial use and zero otherwise

Control

UOU

Dummy variable, equals one if the property’s upper story is for office use and zero otherwise

+

Control

URU

Dummy variable, equals one if the property’s upper story is for residential use and zero otherwise

+

Control

LnINDEX

Private Retail Prices Index (1999 = 100) + (in natural logarithm form)

Control

OTHERS

Number of non-IVS visitors

Control

IVS

Number of visitors under the IVS

+ +

Control

IVS × LnMTR Interaction between IVS and LnMTR



H1

IVS × ACM

Interaction between IVS and ACM

+

H2A and H2B

IVS × AGE

Interaction between IVS and AGE



H3

Rating and Valuation Departments of Hong Kong. Moreover, we manually collected the frontage length with a laser measurer during site visits.

3.6 Results

35

Table 3.2 Descriptive statistics of continuous variables Variable

Median

Max

Min

ACM

20.65

16

54

0

14.83

AGE

32.49

34.21

53.5

0.43

11.45

3.7

22.76

0

138.5

344.6

79.4

FRON

Mean

4.23

INDEX

152.87

IVS

358.93

Std. Dev

3.05 60.29

627.98

1786.25

0

LnMTR

5.44

5.35

6.55

2.64

0.67

LnMALL

5.66

5.67

7

2.4

0.68

PRICE OTHERS SIZE

122.24 1300.73 59.41

463.99

73.34

787.41

2.41

128.89

1353.82

2175.31

427.25

390.36

45.06

656.08

5.02

58.5

Note Prices are at 1999 constant price levels

3.6 Results A pairwise correlation matrix was first calculated to showcase the correlation between the independent variables. Results (not shown here) show that collinearity is not an issue for the data. Table 3.3 lists the OLS estimation results. Most of the variables perform as expected. The model can explain approximately 56% of the variation in retail property prices. The goodness of fit value is acceptable for the limited sample size (N = 580). A Moran’s I test is undertaken for testing the spatial autocorrelation in the data. Results illustrate that the spatial autocorrelation is significant (Moran’s I value = 0.128, p-value < 0.001) and thus reject the null hypothesis (no spatial dependence exists). The results are highly consistent with the majority of, though not all, hedonic studies. As such, the traditional hedonic pricing model that fails to incorporate spatial autocorrelation is deemed to be inappropriate as it produces biased results, which should be interpreted with caution. We opt for spatial econometric models in subsequent analysis. After comparing the performance of the SDM and the other two basic spatial econometric models (i.e., the SLM and the SEM), we find that the SDM performs best, which concurs with the existing literature. Table 3.4 shows coefficient estimates of the SDM. Results illustrate that the SDM fit the data modestly better than the traditional hedonic pricing model (see Table 3.3), or that directly incorporating spatial effects increases the explanatory power of the model. By comparing Tables 3.3 and 3.4, we find that OLS regression is biased and may either overestimate or underestimate the coefficients associated with independent variables. Furthermore, the spatial autoregressive parameter (ρ) is significant at the 1% level and has a positive sign. This result indicates the presence of spatial autocorrelation and agrees with a priori expectations.

36 Table 3.3 Coefficient estimates of the OLS model

3 Time-Varying Impact of Metro … Variable

Coefficient

t-statistic

p-value

AGE

−0.0002

−0.041

0.968

SIZE

0.0151***

11.126

0.000

SIZE2

−0.00002***

−7.008

0.000

FRON

−0.7949**

2.175

0.030

LnMTR

−0.1115***

−6.379

0.000

LnMALL

0.0511

−0.893

0.372

CORN

−0.0008**

2.282

0.023

ACM

0.0352***

5.102

0.000

ACM2

0.2395***

−4.215

0.000

UCU

−0.3118*

−1.896

0.059

UOU

0.3963

1.506

0.133

URU

−0.0819

−0.941

0.347

LnINDEX

−0.0461

−0.257

0.797

OTHERS

0.2116

1.173

0.241

IVS

−1.1021

−1.394

0.164

IVS × LnMTR

−0.0032

−0.456

0.649

IVS × ACM

0.2297*

1.864

0.063

IVS × AGE

0.0057

0.904

0.366

Constant

14.5565***

13.023

0.000

Performance statistic R2

0.569

Adjusted R2

0.555

Number of observations

580

Note *** Significant at the 1% level. ** Significant at the 5% level. * Significant at the 10% level

Table 3.5 reveals direct, indirect (or spillover), and total effects of hedonic variables. A total of 10 variables (e.g., AGE, LnMTR, and LnMALL) have spatial spillover effects on retail property prices. This finding supports the notion that the price of a property is affected by the prices and characteristics of nearby properties. The following interpretations are simply based on total effects. “AGE,” “SIZE,” and “FRON” are classified as structural attributes. The price effect of “SIZE” is positive and significant, whereas that of “SIZE2 ” is negative and significant. This observation shows that an inverted-U (non-linear) relationship exists between size and property prices (in natural logarithm form). Moreover, the price effect of “FRON” is insignificant in this empirical study. A possible reason for this outcome is the limited sample size (N = 580). For locational attributes, the price effect of “LnMTR” is insignificant at the 10% level, indicating that metro accessibility is too weak to shape (or determine) retail property prices before the implementation of the IVS. In addition, the price effect

3.6 Results

37

Table 3.4 Coefficient estimates of the SDM Variable

Coefficient

t-statistic

Variable

Coefficient

AGE

−0.0041

−1.300

W-AGE

0.0302***

SIZE

0.0147***

2.703

W-SIZE

0.0055***

SIZE2

0.0000

−0.012

W-SIZE2

FRON

0.0281***

414.103

W-FRON

0.0358***

LnMTR

−0.5319***

−13.996

W-LnMTR

1.2102***

13.298

LnMALL

0.1117***

2.708

W-LnMALL

−0.9153***

−9.717

CORN

0.2259***

13.658

W-CORN

0.5342***

12.462

ACM

0.0212***

4.598

W-ACM

0.0370***

ACM2

−0.0003

−0.165

W-ACM2

−0.0004

−0.148

UCU

0.0413***

3.944

W-UCU

−0.7653***

−8.900

UOU

0.7428***

3.592

W-UOU

−1.9626***

−3.782

URU

−0.1577

−1.500

W-URU

0.6691***

8.676

LnINDEX

−0.0397***

−6.641

W-LnINDEX

0.4750***

3.542

OTHERS

0.1896***

5.697

W-OTHERS

−0.0695***

−4.316

IVS

−1.2468***

−10.602

W-IVS

14.1643***

26.942

IVS × LnMTR

0.2043***

90.737

W-IVS × LnMTR

−1.9169***

−181.821

IVS × ACM

0.0103***

10.276

W-IVS × ACM

−0.0490***

−4.043

IVS × AGE

0.0035

1.202

W-IVS × AGE

−0.0960***

− 14.666

ρ

0.2780***

16.383

Constant

−0.0001

3.915***

t-statistic 8.698 2.801 −0.032 141.582

3.207

85.369

Performance statistic R2

0.662

Adjusted R2

0.639

Number of observations

580

Note *** Significant at the 1% level. ** Significant at the 5% level. * Significant at the 10% level

of “LnMALL” is negative and significant at the 10% level. This result means that accessibility to shopping malls is positively associated with retail property prices. Furthermore, the price effect of “CORN” is positive and significantly different from zero at the 1% level. This observation implies that the street corner location is positively correlated with retail property prices, consistent with a priori expectations and the existing literature (Nase, 2013). A possible explanation is that the street corner location can attract more spotlights from potential customers (Nase et al., 2013).

38

3 Time-Varying Impact of Metro …

Table 3.5 Direct, indirect, and total price effect estimates of the SDM Variable

Direct effect (t-statistic)

Indirect effect (t-statistic)

Total effect (t-statistic)

AGE

−0.0039 (−0.885)

0.0411* (1.821)

0.0372* (1.668)

SIZE

0.0146*** (11.863)

0.0129 (1.143)

0.0276** (2.461)

SIZE2

0.0000***

−0.0001**

−0.0001*** (−2.908)

FRON

0.0279*

LnMTR

−0.5238***

LnMALL

0.1113 (0.674)

−1.2839* (−1.805)

−1.1726* (−1.823)

CORN

0.2320** (2.503)

0.8208 (1.438)

1.0528* (1.789)

ACM

(−7.558)

(1.833) (−3.516)

(−2.443)

0.0670 (0.741) 1.5536*

(1.836)

0.0949 (1.063) 1.0298 (1.262)

0.0214 (1.475)

0.0611 (1.240)

0.0825* (1.787)

ACM2

−0.0003 (−0.986)

−0.0008 (−0.937)

−0.0010 (−1.430)

UCU

0.0456 (0.250)

−1.0706** (−1.968)

−1.0251* (−1.801)

UOU

0.7442*** (2.899)

−2.7011* (−1.657)

−1.9569 (−1.184)

URU

−0.1609 (−1.538)

0.8949** (2.084)

0.7340* (1.793)

LnINDEX

−0.0417 (−0.261)

0.6126 (0.555)

0.5710 (0.504)

OTHERS

0.1914 (1.195)

−0.1213 (−0.086)

0.0701 (0.049)

IVS

−1.1895* (−1.705)

19.8424** (2.498)

18.6529** (2.298)

IVS × LnMTR

0.1960* (1.775)

−2.6588** (−2.237)

−2.4628** (−2.011)

IVS × ACM

0.0102*

−0.0649 (−1.203)

−0.0547 (−0.991)

IVS × AGE

0.0032 (0.474)

−0.1362*** (−3.480)

−0.1330*** (−3.399)

(1.792)

Note *** Significant at the 1% level. ** Significant at the 5% level. * Significant at the 10% level

For neighborhood and other variables, “ACM,” which reflects the cumulative opportunities of accommodation facilities, has a positive price effect before the implementation of the IVS, whereas “ACM2 ” has an insignificant price effect. This outcome is reasonable and indicates a positive linear relationship between accessibility to accommodation facilities and retail property prices (in natural logarithm form). It can be explained by the fact that retail properties with better accessibility to accommodation facilities are more likely to be patronized by tourist shoppers. Moreover, regarding the vertical neighborhood use variables, “UCU” and “URU” have significant negative and positive price effects, respectively. However, “UOU” is insignificant at the 10% level. Furthermore, “IVS” is significant at the 5% level, and its price effect holds a positive sign. This finding confirms the significant positive price effect of the IVS (which introduces numerous Mainland Chinese tourist shoppers) in Causeway Bay and is consistent with Li et al. (2018). The interpretation of the three interaction terms that are directly related to hypothesis testing is of predominant interest in this study. The summary of the corresponding results is shown in Table 3.6. First, the price effect of “IVS × LnMTR” is negative and significant at the 5% level, which verifies H1. This outcome indicates that after the implementation of the IVS, metro accessibility has been more valued and is associated with higher retail property prices. This finding and is consistent with previous

3.6 Results Table 3.6 Summary of the three sets of hypotheses

39 Theoretical Background

Economic concern

Psychological implication

Hypothesis

H1

H2A and H2B

H3

Hedonic variable

IVS × LnMTR

IVS × ACM

IVS × AGE

Expected sign −

+/insignificant



Test result

Reject H2A and Confirm confirm H2B

Confirm

research. Second, the price effect of “IVS × ACM” is insignificantly different from zero, which contrasts H2A but confirms H2B. This means that the economic value of accessibility to accommodation facilities is not significantly altered by the implementation of the IVS. Last, the price effect of “IVS × AGE” is negative and significant at the 5% level, which verifies H3 and agrees with Li et al. (2018).

3.7 Conclusions and Discussion This study investigates the impact of the increase in tourist shoppers on the prices of retail properties in the tourist precinct of Hong Kong (a typical shopping destination where tourists allocate a high budget for shopping during their trips). Based on previous studies and relevant theories, three sets of hypotheses related to economic and psychological concerns are developed. This study makes use of the implementation of the policy IVS in Hong Kong in 2003 and the transaction records of groundfloor retail properties in Causeway Bay during 1993–2011 to test these hypotheses. Notably, due to the presence of spatial autocorrelation, conventional hedonic price models may lead to biased results. This study solves this problem by employing a widely-used spatial econometric technique, namely the SDM. Our findings are listed as follows. (1) The implementation of the IVS has a positive impact on retail property prices. (2) The implicit price of metro accessibility increases after the implementation of the IVS. (3) The implicit price of accessibility to accommodation facilities is not significantly altered by the implementation of the IVS. (4) Age has a larger negative price effect after the implementation of the IVS. This study advances our understanding of the interaction between tourism demand and retail property prices in a shopping destination and enriches or supplements the existing literature on this topic. Moreover, our empirical results can have significant practical implications. Tourism policy-makers and practitioners can improve the physical environment of shopping spaces, which can greatly attract more potential consumers. Public sectors can organize regular refurbishment in the core areas of a tourist destination to enhance the attractiveness of the destination. Retail practitioners can invest in newer retail shops to pursue higher returns. These implications

40

3 Time-Varying Impact of Metro …

are believed to be applied to other shopping destinations with an increasing volume of tourists. The study makes a small step towards the exploration of possible interactions between the tourism industry and the retail real estate market. Indeed, several future research directions exist. We point out two: (1) analyzing how changes in the localtourist shopper mix affect the dynamics of the rental and vacancy adjustment in the retail space market using time series or panel data; and (2) examining the determination of the equilibrium retail property price as a result of the change in macro-economic variables of the tourism industry. There are a few research limitations. First, to test the three sets of hypotheses, this study focuses on a small geographical area (i.e., Causeway Bay) instead of the whole city or a large area (e.g., Hong Kong Island and Kowloon). The area is carefully selected by the authors. This economically rigorous approach helps control for numerous locational and neighborhood attributes (e.g., accessibility to Luohu Port, Lok Ma Chau Port, and Shenzhen Bay Port) and thus primarily relieves, though definitely does not eliminate, a much-derided problem of hedonic pricing, namely missing variable bias. Admittedly, it is far from perfect and suffers from the following two distinct shortcomings: (1) generalization or transferability of the results and (2) limited sample size. The number of observations used for analysis (N = 580) is much less than that for a wider area, which may distort the results. Therefore, we suggest that more empirical studies on this topic should be conducted, which is indispensable to reach stronger conclusions. Second, this study only links tourism demand to the retail trade. Understandably, the wholesale trade is also expected to be affected by tourism demand. Analyzing their interaction is interesting and fascinating, but it cannot be completed by this study because the wholesale trade of Hong Kong is normally not conducted in Causeway Bay. We suggest that it should be explored in upcoming research.

References Benjamin, J. D., Boyle, G. W., & Sirmans, C. F. (1990). Retail leasing: The determinants of shopping center rents. Real Estate Economics, 18(3), 302–312. Choi, M., Law, R., & Heo, C. Y. (2016). Shopping destinations and trust–tourist attitudes: Scale development and validation. Tourism Management, 54, 490–501. Jayantha, W. M., & Yung, E. H. K. (2018). Effect of revitalisation of historic buildings on retail shop values in urban renewal: An empirical analysis. Sustainability, 10(5), 1418. Li, L. H., Cheung, K. S., & Han, S. Y. (2018). The impacts of cross-border tourists on local retail property market: An empirical analysis of Hong Kong. Journal of Property Research, 35(3), 252–270. Lindquist, J. D. (1974). Meaning of image-survey of empirical and hypothetical evidence. Journal of Retailing, 50(4), 29–38. Manski, C. F. (1993). Identification of endogenous social effects: The reflection problem. The Review of Economic Studies, 60(3), 531–542. Meng, F., & Xu, Y. (2012). Tourism shopping behavior: Planned, impulsive, or experiential? International Journal of Culture, Tourism and Hospitality Research, 6, 250–265.

References

41

Nase, I., Berry, J., & Adair, A. (2013). Hedonic modelling of high street retail properties: A quality design perspective. Journal of Property Investment & Finance, 31(2), 160–178. Xu, W. A., & Yang, L. (2019). Evaluating the urban land use plan with transit accessibility. Sustainable Cities and Society, 45, 474–485. Xu, Y., & McGehee, N. G. (2012). Shopping behavior of Chinese tourists visiting the United States: Letting the shoppers do the talking. Tourism Management, 33(2), 427–430. Yang, L., Zhou, J., Shyr, O. F., & Huo, D. (2019). Does bus accessibility affect property prices? Cities, 84, 56–65. Yang, L., Chau, K. W., Lu, Y., Cui, X., Meng, F., & Wang, X. (2020). Locale-varying relationships between tourism development and retail property prices in a shopping destination. International Journal of Strategic Property Management, 24(5), 323–334.

Chapter 4

Accessibility and Proximity Effects of Bus Rapid Transit on Property Prices

4.1 Introduction More and more cities have been vigorously taking policy measures to create a livable environment for citizens to meet the needs of environmental protection in the era of sustainable development. Considering that private cars emit a great amount of vehicle exhaust and aggravate traffic congestion, public transport, such as urban rail transit and bus rapid transit (BRT), is gaining universal acceptance worldwide and attracting substantial attention from, inter alia, governments and researchers because of its economic feasibility, cost-effectiveness, and environmental friendliness. It is especially essential for cities striving for sustainability. BRT, which combines the carrying capacity of the metro and the flexibility and low costs of conventional bus transit, has gained increasing popularity around the world in recent years, particularly in developing countries with capital resource scarcity (Bocarejo et al., 2013). BRT could help establish an accessible and affordable public transit system in densely inhabited cities and meet the needs of sustainable development in the era of environmental protection (Rodríguez et al., 2016). As of 2018, BRT has been implemented in 170 cities worldwide, with a total length of more than 5000 km and a daily ridership of over 33 million. Latin America and Asia are the two main continents in which BRT is gaining popularity, with more than 89% of the world’s ridership and 66% of the world’s mileage. Curitiba (Brazil) constructed a famous BRT system in 1974. Since then, quite a few cities (e.g., Los Angeles and Quebec) in the West have implemented BRT schemes as a supplement to their residents’ travel mode choice. In 1999, BRT was introduced to urban China where population density is high and citizens show a high preference to live in downtown areas: the first proto-BRT scheme (though not a fullyfledged BRT scheme) was launched in Kunming. Subsequently, mainly owing to the recommendation of the Ministry of Construction, BRT systems have been successfully implemented in several Chinese cities (e.g., Beijing, Guangzhou, Chengdu, and Xiamen) as a promising strategy for relieving traffic congestion problems. As of the end of 2017, 21 Chinese cities have owned a BRT system with a total mileage of © The Author(s), under exclusive license to Springer Nature Singapore Pte Ltd. 2021 L. Yang, Property Price Impacts of Environment-Friendly Transport Accessibility in Chinese Cities, https://doi.org/10.1007/978-981-16-8833-1_4

43

44

4 Accessibility and Proximity Effects …

3424.5 km and a total annual ridership of 2.20 billion. Moreover, based on the plan, the total mileage of BRT in China would reach 5000 km by 2020 (Velásquez et al., 2017). Like most public transit systems (e.g., metro and conventional bus transit), BRT offers station-to-station services rather than door-to-door services Yang et al. (2020). Therefore, accessibility to BRT stations affects adjoining dwellers’ travel mode choice. Thus, BRT has the potential to provide accessibility-based benefits and affect prices of nearby properties because, generally, residents are willing to pay for such a virtually uniformly favorable and desired property attribute (i.e., BRT accessibility) (Ingvardson & Nielsen, 2018). Numerous previous studies have concluded that BRT could offer price premiums to properties (e.g., Deng & Nelson, 2013; Deng et al., 2016). By contrast, proximity to the transit corridor may have negative effects on property values because it is often related to air pollution, noise, traffic congestion, and road vibration (Kilpatrick et al., 2007). However, in stark contrast with well-documented accessibility-based benefits of BRT, the proximity-induced nuisances have seldom been the focus of scholarly attention. Thus, it is worthwhile to explore the association between proximity to the BRT corridor and property prices, particularly in the context of China, which has the highest rate of BRT network expansion. Simultaneous estimation of the joint (i.e., accessibility-based and proximity-induced) effects of BRT on property prices would potentially assist local governments to better capture values added by BRT systems and help property purchasers make informed and reasonable decisions. Recently, a multitude of studies on BRT and property prices/rents has elicited growing attention. Many studies were conducted in western countries, including the United States, Colombia, Australia, and Canada. Among them, the BRT system in Bogotá (i.e., TransMilenio) has elicited the greatest interests. By comparison, very limited scholarly attention has been poured into bus-dependent Chinese cities, in which BRT may play an important role in the daily travel of city dwellers. Furthermore, most studies found that accessibility to BRT brings price premiums to adjacent properties (e.g., Mulley et al., 2016; Rodríguez & Mojica, 2009; Rodríguez & Targa, 2004) and confirmed the benefits of BRT in hard currency. By contrast, a few studies argued that the effect was not significant (e.g., Mulley & Tsai, 2017; Zhang & Wang, 2013) or even negative (e.g., Cervero, 2004; Salon et al., 2014). As formerly mentioned, existing studies mainly focus on the accessibility-based benefits of BRT. However, proximity to the BRT corridor may be associated with a multitude of nuisances. Very few studies have simultaneously considered the accessibility-based and proximity-induced effects of BRT. Some exceptions include the work of Rodríguez and Targa (2004) and Munoz-Raskin (2010), which focused on Bogotá. A better understanding of the holistic impact of BRT on property values should be furthered. The association between BRT and property prices has rarely been investigated systematically in urban China. The only two study areas are Beijing and Guangzhou, both of which have a metro system. However, only 32 Chinese cities had metro systems in 2017, and bus-dependent cities (which have high bus patronage but do not

4.1 Introduction

45

have a metro system) constitute the majority of Chinese cities. Thus, based on 16,165 property observations within 2 km from BRT stations in Xiamen (China), we estimate a multitude of hedonic pricing models to investigate the price premiums and discounts offered by the BRT system. Robustness checks ensure the plausibility of this study. Moreover, we calibrate an enhanced model to test the hypothesis that accessibility to BRT exerts a larger benefit in the peripheral area (wherein transportation alternatives are limited) than in the central area. The contributions of this study include: (1) providing the latest available review of the existing literature on BRT and property prices/rents; (2) offering an empirical case in urban China on evaluating the effects of accessibility to BRT stations and proximity to the BRT corridor on property prices; (3) approaching a comprehensive understanding of the price premiums and discounts associated with BRT; (4) investigating whether the effect of BRT accessibility on property prices is different in the central and peripheral areas; and (5) giving an enlightening implementation and feasible benchmark of financing BRT infrastructure and mitigating proximity-induced discounts for the local government. The remainder of this chapter is organized as follows. Section 4.2 gives a brief description of Xiamen and its Xiamen BRT system. Sections 4.3 and 4.4 describe Box-Cox transformation and spatial econometric models, respectively. Section 4.5 introduces the property price data utilized. Section 4.6 showcases the results of hedonic modeling. Section 4.7 draws policy implications. Section 4.8 provides conclusions and indicates future research directions.

4.2 Overview of Xiamen and Its BRT System 4.2.1 Xiamen The study area is Xiamen Island (Fig. 4.1), the city proper of Xiamen (also known as Amoy in Hokkien), a sub-provincial city located in the southeast part of Fujian Province and at the heart of the Western Taiwan Straits Economic Zone (haixia xi’an jingqi qu). Enticingly called the “Garden on the Sea (haishang huayuan)” and “Oriental Hawaii (dongfang xiaweiyi),” Xiamen has been a seaport city since ancient times and one of the major ancestral homes of overseas Chinese. The Kinmen Islands (Taiwan) lie less than 10 km away. Moreover, the city is one of the five separate itemized cities (jihua danlie shi), which enjoy a certain autonomy and are treated separately from the province where they are situated, of China, as well as a major ancestral home of overseas Chinese. The city covers a land area of approximately 1701 km2 and a sea area of approximately 390 km2 , with a permanent resident population of 4.29 million and a GDP of 599.5 billion yuan (as of 2019). The city is made up of Siming, Huli, Xiang’an, Tong’an, Jimei, and Haicang districts. Furthermore, public transport accounts for over 30% of residents’ trips.

46

Fig. 4.1 Location of Xiamen City and Xiamen Island

4 Accessibility and Proximity Effects …

4.2 Overview of Xiamen and Its BRT System

47

Xiamen Island is the political, economic, cultural, and tourist center of the city. With an administrative area of approximately 130 km2 , it is 13.7 km long from north to south and 12.5 km wide from west to east. It is composed of two districts, Siming and Huli. Siming district, from which the city originates, is home to the city government, the inter-city railway station, the city center of Xiamen (i.e., Zhongshan Road area), and the best education, healthcare, and cultural facilities of the city. Thus, it has always been regarded as the central area of Xiamen Island (and of the whole city). Therefore, Huli district is perceived as the peripheral area of Xiamen Island. The city opened its metro system (only 1 line) on the final day of 2017, approximately ten months after the property price data collection. The metro line mainly connects the Xiamen North Railway Station and Zhenhai Road (adjacent to Kulangsu Island). A common shortcoming of empirical data-modeling methodologies, such as the hedonic price modeling, is the omitted variable bias. An effective approach to circumventing this problem is focusing on a narrow geographic area where a host of confounding variables can be properly controlled (Brasington, 2003). The scale and geographical settings of the island make it a tractable laboratory for conducting this hedonic research.

4.2.2 Xiamen BRT System The BRT system in Xiamen, owned by the Xiamen BRT Operation Company, officially entered operation with elevated, dedicated, and exclusive lanes (or running way, right-of-way, busway) in 2008 (Fig. 4.2). Afterward, it gradually entered a mature operational stage and became an indispensable and cheap transportation mode for local residents with a daily ridership of 300,000 as of 2019. Owing to the complete separation from other traffics accrued from the spectacular elevated running way, the city center peak-hour speed is extremely high (31.6 km/h), considerably greater than most counterparts in Chinese cities, such as Beijing (16–20 km/h), Guangzhou (18 km/h), and Zhongshan (26 km/h). The key performance indicators of the three BRT lines in Xiamen Island are summarized in Table 4.1. The service time is from 5:45 am to 10:50 pm. The BRT adopts a distance-based ticket pricing structure and an off-board fare collection method. The initial price of the BRT is 1 yuan (1 yuan is equivalent to approximately 0.145 US dollar) within the first 5 km, and the cumulative price will increase by 0.15 yuan for each additional 1 km. The maximum one-way price is 4 yuan. In most cities (e.g., Bogotá and Guangzhou), the BRT lane is typically aligned to the center of the road, and BRT operates on the same plane with other travel modes (e.g., private car and conventional bus transit), thereby degrading pedestrian connection. By contrast, BRT lanes in Xiamen are elevated (see Fig. 4.2). The elevation of the BRT right-of-way partially blocks out the sights of bilateral pedestrians and drivers and spoils the urban landscape. Other than air pollution, noise, traffic congestion, and road vibration (which are associated with most, though not all, BRT

48

4 Accessibility and Proximity Effects …

Fig. 4.2 The BRT system on Xiamen Island

corridors), the elevation brings about additional nuisances, namely loss of privacy and shading. The elevated bus lanes are only for BRT use. This feature differentiates Xiamen Island’s BRT system from those around the world (e.g., in Bogotá, Beijing, and Chengdu) that share the same operation plane (despite dedicated lanes) with other

4.2 Overview of Xiamen and Its BRT System Table 4.1 Key performance indicators of the Xiamen BRT

49

Indicator

Line 1

Line 2

Line 3

Year of commencing

2008

2008

2008

Corridor service

Trunk lines

Trunk lines

Trunk lines

Corridor length (km)

28.20

10.00

10.70

No. of stations

26

32

15

Station spacing, corridor (m)

1340.0

2500.0

710.0

All

All

Pre-boarding fare All collection

Source brtdata.org/location/asia/china/xiamen

transportation modes (e.g., cars, conventional bus transit, walking, and bike-sharing). In Xiamen Island, for the most part, BRT buses are exclusively permitted to be operated on the elevated BRT lanes (Fig. 4.2a) and do not run away from the BRT system infrastructure (Wright & Hook, 2007; Zhang et al., 2020); and all the other transportation modes are operated under the lanes (see Fig. 4.2b). In other words, the corridor access is strictly limited to BRT buses. Therefore, Xiamen Island’s BRT system is a closed system, similar to Bogotá’s TransMilenio and Curitiba’s Rede Integrada de Transporte. Moreover, Xiamen Island’s BRT system uses a single vehicle to operate throughout the region, and terminal transfer, which is applicable to the “trunk-feeder” configuration, is not involved. To sum up, the studied BRT system is a closed system with a “direct services” configuration, offering “highly flexible operating conditions and a high-quality service” (Wright & Hook, 2007, p. 4). Considering the importance of BRT in Xiamen, we can expect that BRT has notable impacts on the value of adjacent properties.

4.3 Snapshot of Box-Cox Transformation Given the lack of a priori economic theory dictating a correct specification of the hedonic pricing model, the validity of any prespecified model can indeed be questioned and challenged. As such, more flexible models can be derived for consideration using the Box-Cox transformation. The Box-Cox transformation is a typical non-linear regression technique that employs an iteration process that maximizes the model log-likelihood. It can account for non-linearity in model parameters and renders the residuals more closely normal and less heteroskedastic. Box-Cox models have various functional forms, such as the simple left-hand-side, simple right-hand-side, simple both-side, and separate both-side models. The technique has been employed widely in numerous hedonic studies (Yang et al., 2018, 2019).

50

4 Accessibility and Proximity Effects …

The separate both-side model is generally the most general specification of the Box-Cox transformed model, which can be formulated as follows. Y (θ) = X (λ) β + ε, where Y (θ) = (Y θ − 1)/θ for θ = 0, X (λ) = (X λ − 1)/λ for λ = 0, whereas Y (θ) = ln Y for θ = 0, X (λ) = ln X for λ = 0; Y is the property price; X is relevant attributes of the property; β is coefficients of independent variables; and E is a residual that captures all unmeasured factors. θ and λ is normally called the left-side parameter and right-side parameter, respectively. In contrast to the separate both-side model that transforms left- and right-side variables using different parameters, the left-hand-side model only transforms the left-side variable, and the simple both-side model restricts that the left-side parameter is equivalent to the right-side parameter (i.e., θ = λ).

4.4 Snapshot of Spatial Econometric Models Given that the standard hedonic pricing model fails to account for the presence of spatial dependence (spatial autocorrelation), defined as “the co-variation of variables within a geo-space” (Li et al., 2011, p. 443), the employment of spatial econometric models becomes an obvious trend in real estate valuation and econometrics research. In this vein, spatial econometric methods can be regarded as advanced versions of the hedonic pricing model. The main motivation of spatial modeling is to consider “near and related things.” Thus, if spatial autocorrelation is present, failing to use spatial regression models will lead to biased and inconsistent coefficients. Employing spatial econometric models improves the reliability of our findings and conclusions. To the best of our knowledge, using traditional OLS regression for spatial data is no longer as acceptable as before, and academia has increasingly refrained from employing the OLS and traditional hedonic modeling, largely as a result of the development of spatial econometric methods that can be estimated easily with growing computer power. An ongoing diffusion of Geographic Information Systems has also served to considerably ease the process of incorporating spatial dimensions into the analysis, thereby paving the way for the rapid development of spatial modeling. Krause and Bitter (2012) reviewed the existing literature on applications of spatial econometric models in the real estate market. The spatial lag model (SLM, a.k.a. spatial autoregressive model) and spatial error model (a.k.a. spatial autoregressive error model) are two well-known basic methods addressing the presence of spatial dependence. They have been widely applied in numerous hedonic studies. The SLM addresses the interaction effects among the dependent variable. It assumes that a dependent variable is affected by both independent variables and

4.4 Snapshot of Spatial Econometric Models

51

the dependent variable of nearby observations. The model can be formulated as Y = ρW Y + Xβ + ε, where W is a spatial weight matrix, which specifies the structure of the spatial relationship among observations and the information on which observations are considered neighbors and how their values are related to one another W Y denotes the spatially lagged dependent variable; and ρ is the spatial autocorrelation parameter (normally between 0 and 1). The spatial weight matrix is based on either contiguity or distance, and its dimensions (N × N) are based on the sample size (N). When ρ = 0, the SLM is simplified as the standard hedonic model. ρ and β are the parameters to be estimated simultaneously. The SEM addresses spatial autocorrelation in the residuals and can be formulated as Y = Xβ + ε, ε = λW ε + u, where u is a residual that is assumed to have no correlation with other observations’ error terms; and λ is the spatial autocorrelation parameter. When λ = 0, the SEM is simplified as the standard hedonic model. λ and β are the parameters to be estimated simultaneously. The SLM and SEM can also be written as Y = (I −ρW )−1 Xβ +(I −ρW )−1 ε and Y = Xβ + (I − λW )−1 u, respectively. So we can find that the estimated coefficients of the SEM can be interpreted in the usual way as marginal effects, while those of the SLM cannot be directly interpreted. The reason is that the marginal effect on property price revealed by the SLM is the summation of both a direct effect (arising from the change in the amount of the attribute) and an indirect/induced/spillover/feedback effect (arising from the marginal changes related to the neighbors’ values).

4.5 Property Price Data A dataset of secondhand residential properties (or houses) throughout Xiamen Island was extracted from a well-known property website in China (Sofang.com) in March 2017. The sample includes the asking price and location information of 22,586 properties. A more detailed description of this dataset can be found in Yang et al. (2018). Additionally, if BRT is too far away from home, dwellers will not consider BRT when selecting travel modes. In other words, the impact of BRT on property prices only exists within a certain distance. Similar to most, if not all, existing studies that confined the sample to BRT-adjacent areas (e.g., Rodríguez & Targa, 2004; Salon et al., 2014), the sample of this study is confined to 16,165 properties within 2 km from any BRT station.

52

4 Accessibility and Proximity Effects …

Table 4.2 displays the definitions and summary statistics of variables used for hedonic modeling. In this study, the dependent variable is property price. A wide array of property attributes, such as gross floor area, age, and building height, is controlled for in the hedonic pricing models. Moreover, with the aim of imparting higher flexibility, our models include dummy variables rather than a positive integer variable to represent the contribution of the number of bedrooms. Moreover, a series of location and neighborhood variables is introduced to address the effects of amenities and public utilities, such as the city center, the sea, waterbody, and bus stations. Specifically, accessibility to BRT is measured by distance to the nearest BRT station, whereas proximity to the BRT corridor is quantified by dummy variables (