Economic Analysis of Housing Policy in Japan: Policy Concerning Housing Quality (New Frontiers in Regional Science: Asian Perspectives, 64) 9811949247, 9789811949241

Citation preview

New Frontiers in Regional Science: Asian Perspectives 64

Keiko Nosse Hirono

Economic Analysis of Housing Policy in Japan Policy Concerning Housing Quality

New Frontiers in Regional Science: Asian Perspectives Volume 64

Editor-in-Chief Yoshiro Higano, University of Tsukuba, Tsukuba, Ibaraki, Japan

This series is a constellation of works by scholars in the field of regional science and in related disciplines specifically focusing on dynamism in Asia. Asia is the most dynamic part of the world. Japan, Korea, Taiwan, and Singapore experienced rapid and miracle economic growth in the 1970s. Malaysia, Indonesia, and Thailand followed in the 1980s. China, India, and Vietnam are now rising countries in Asia and are even leading the world economy. Due to their rapid economic development and growth, Asian countries continue to face a variety of urgent issues including regional and institutional unbalanced growth, environmental problems, poverty amidst prosperity, an ageing society, the collapse of the bubble economy, and deflation, among others. Asian countries are diversified as they have their own cultural, historical, and geographical as well as political conditions. Due to this fact, scholars specializing in regional science as an inter- and multi-discipline have taken leading roles in providing mitigating policy proposals based on robust interdisciplinary analysis of multifaceted regional issues and subjects in Asia. This series not only will present unique research results from Asia that are unfamiliar in other parts of the world because of language barriers, but also will publish advanced research results from those regions that have focused on regional and urban issues in Asia from different perspectives. The series aims to expand the frontiers of regional science through diffusion of intrinsically developed and advanced modern regional science methodologies in Asia and other areas of the world. Readers will be inspired to realize that regional and urban issues in the world are so vast that their established methodologies still have space for development and refinement, and to understand the importance of the interdisciplinary and multidisciplinary approach that is inherent in regional science for analyzing and resolving urgent regional and urban issues in Asia. Topics under consideration in this series include the theory of social cost and benefit analysis and criteria of public investments, socio-economic vulnerability against disasters, food security and policy, agro-food systems in China, industrial clustering in Asia, comprehensive management of water environment and resources in a river basin, the international trade bloc and food security, migration and labor market in Asia, land policy and local property tax, Information and Communication Technology planning, consumer “shop-around” movements, and regeneration of downtowns, among others. Researchers who are interested in publishing their books in this Series should obtain a proposal form from Yoshiro Higano (Editor in Chief, [email protected]) and return the completed form to him.

Keiko Nosse Hirono

Economic Analysis of Housing Policy in Japan Policy Concerning Housing Quality

Keiko Nosse Hirono College of Economics Nihon University Chiyoda-ku, Tokyo, Japan

ISSN 2199-5974 ISSN 2199-5982 (electronic) New Frontiers in Regional Science: Asian Perspectives ISBN 978-981-19-4924-1 ISBN 978-981-19-4925-8 (eBook) https://doi.org/10.1007/978-981-19-4925-8 © Springer Nature Singapore Pte Ltd. 2022, corrected publication 2022 This work is subject to copyright. All rights are reserved 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

To My Family

Preface

This book aims to study housing policy in terms of the housing quality, theoretically and empirically, and to provide policy proposals. The background and main contents of this book are as follows. Firstly, the book shows Japanese government’s successful endeavors in increasing the volume of houses. According to Ministry of Internal Affairs and Communications’ 2018 Housing Land Survey, in Japan, the numbers of vacant and total houses account for 8.46 million and 62.41 million, respectively. This indicates that the proportion of vacant houses accounts for 13.6% of the total houses in Japan and that every household has an average of 1.16 housing units. Japan achieved this volume through its interest rate policy, as demonstrated by the Vector Autoregression analysis in this book. The country’s housing policy is currently focusing on the housing quality. To the best of my knowledge, this is the first book to analyze and propose housing monetary and subsidy policies to improve the housing quality, and thereby increase the number of barrier-free and earthquake-resistant houses. Secondly, the research on housing price movements, my former work under Professor Takatoshi Ito in the Bank of Japan, required the creation of a housing price index. The research showed the inadequacy of the average prices of houses with different attributes in developing a housing price index. Hence, I built a qualityadjusted housing price index. Thirdly, there exist the limitations of the conventional housing evaluation methods comprising the cost, sales comparison, and profit return methods. The cost method considers only the suppliers’ side and does not necessarily reflect the actual situation of the housing market, including that of the buyers. The sales comparison method cannot be used in all the cases, given the difficulty in finding appropriate houses in a region for comparison. The profit return method does not provide an accurate estimation of future profits. However, the hedonic method presents a valuation of houses based on market clearing prices. Therefore, in this book, I construct an economic model and develop a method for an accurate valuation of houses at market clearing prices, by removing analytical errors and using the hedonic approach. vii

viii

Preface

To the best of my knowledge, this is the first book to develop an appropriate housing valuation method without errors based on the hedonic price function and the housing quality. The valuation method can facilitate the calculation of the real theoretical prices of houses, without analytical errors. This would require information disclosure about houses, and hence this book proposes a housing policy focusing on such disclosures. In addition, this book develops a method to estimate the hedonic price and rent indexes on the basis of the housing quality in Japan. This book has several features. Firstly, it presents methods to improve the housing quality. Housing is closely related to life. Thus, it is crucial to improve the housing quality, which also contributes toward increasing housing amenities in Japan. This comprises one of the important public policy targets in Japan. Secondly, it proposes policymakers and researchers several measures to improve housing policies concerning housing quality. Specifically, the appropriate housing valuation method presented in this book evaluates houses at market clearing prices, corrects analytical errors, and leads to pareto optimal allocation, thereby contributing toward improving the housing markets in different regions. This book recommends policymakers to promote the use of this housing valuation method using the hedonic function and to disclose information of houses. The measures to construct the housing price and rent indices can help policymakers and researchers determine or study the housing market’s movements. The book proposes a method of funding rental housing with a securities investor and introducing rent subsidies to accelerate the construction of rental housing apt for the elderly. It also presents measures to estimate the amount of subsidy to increase the number of earthquake-proof houses. Thirdly, the propositions in Chap. 1 of this book can help companies benefit from the use of the hedonic housing valuation and hedonic price and rent indexes. Fourthly, the measures presented in this book are based on theoretical and empirical analyses of economics. These improvement measures are also practical and correspond to the actual economy. Therefore, several studies forming the bases of chapters in this book are listed in scholarly search engines, such as Econlit or SSRN, as well as in the reference of textbooks for practitioners or Justia. Fifthly, there are few economic analyses on housing quality. This gap is covered by presenting new propositions and findings related to housing quality in each chapter of this book. For example, the book introduces the hedonic housing function for the valuation of housing and hedonic housing price index in Japan. The book also shows the relationship between the amount of homeownership investment and differences in information, and housing valuation at market price by removing errors and facilitating government information disclosure. It further presents empirical findings of the effects of the Government Housing Loan Corporation’s housing loan policy on housing starts in Japan. Finally, it presents ways to estimate the subsidy required to improve the earthquake-proof conversion of rental housing and policies to accelerate barrier-free rental housing. Hence, the propositions in this book can contribute toward solving housing problems in aging countries requiring barrier-free housing, earthquake-prone countries, and countries with high house vacancy due to declining population.

Preface

ix

The layout of this book is as follows. Chapter 1 develops the method of housing valuation using the hedonic housing function and estimation of the hedonic price and rent function and shows their use. Chapter 2 demonstrates the extent to which differences in home ownership investments are caused by differences in information about the property. This finding shows the need to develop a method for appropriate housing valuation without errors in Chap. 3. Chapter 3 also shows that government information disclosure is necessary to achieve pareto optimal allocation in the housing market. Chapter 4 is devoted to a time series analysis of the effect of the low interest rate policy of the Government Housing Loan Corporation on housing starts. Chapter 5 analyzes the subsidy required to make rental houses earthquakeproof, where the quality of the house reflects its level of earthquake resistance. Chapter 6 explores housing policies supporting an aging society. Tokyo, Japan

Keiko Nosse Hirono

The original version of the book has been revised. A correction to this book can be found at https://doi.org/10.1007/978-981-19-4925-8_7

Acknowledgments

I would like to express gratitude to my former academic advisor at the Faculty of Economics of Tokyo University, Dr. Koichi Hamada, who is currently a professor emeritus, Department of Economics, Yale University. I would also like to thank my former supervisor at the Department of Economics, Graduate School of Hitotsubashi University, Dr. Juro Teranishi, who is currently a professor emeritus, the Department of Economics, Hitotsubashi University. I am grateful to Professor Takatoshi Ito, Columbia University, for the joint research project at the Bank of Japan, which gave this book its starting point. I am indebted to the supervisor of my doctoral thesis, Professor Masayuki Nakagawa, the Department of Economics, Nihon University. I am deeply grateful to the late Professor Keiichi Tanaka, Nihon University, for contributing toward the foundation of this research in the field of housing policy. I am indebted to Professor Kiyohiko Nishimura, Professor Yoshiro Tutui, Professor Ryuzo Sato, Professor Yasuhiro Sakai, Professor Yuzo Honda, Professor Taiji Asami, Professor Toshihiro Ihori, and Professor Akira Yokoyama for their helpful comments at academic conferences and workshops. I would like to thank Professor Li Ke and Professor Tetsuro Shimamoto for their suggestions and the research stimuli. This research was partially supported by the Japanese Government (Ministry of Education, Culture, Sports, Science and Technology), the Grants-in-aid for Scientific Research (A) (15203015, Keiko Hirono). I am very grateful to Professor Higano, the editor-in-chief of the series New Frontiers in Regional Science: Asian Perspective, for giving me the opportunity to publish this book in the series. I am thankful to Professor Makoto Tawada for his assistance and to Mr. Yutaka Hirachi and Mr. Selvakumar Rajendran of Springer for their cooperation and patience.

xi

Contents

1

2

Estimation and Applications of the Hedonic Function of Housing . . 1.1 Introduction . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 1.2 Process of Research on Housing Valuation Using the Hedonic Function . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 1.3 Literature Survey . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 1.4 The Hedonic Method . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 1.5 Method for Estimating the Housing Price and Marginal Valuation of Attributes . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 1.6 The Hedonic Price Index . . . . . . . . . . . . . . . . . . . . . . . . . . . . 1.7 Examples of Goods . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 1.8 Estimation of Hedonic Price Index . . . . . . . . . . . . . . . . . . . . . 1.9 Examples of Hedonic Housing Price Index . . . . . . . . . . . . . . . 1.10 Advantages and Applications of Hedonic Housing Price and Rent Function . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 1.11 Use of Hedonic Price and Rent Indices . . . . . . . . . . . . . . . . . . 1.12 Conclusion . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . References . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . Imperfect Information and Varying Homeownership Investments: Utilizing Deviation from the Rational Expectations Price . . . . . . . . 2.1 Introduction . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 2.2 The Model . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 2.3 Data . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 2.4 Empirical Findings . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 2.4.1 Calculating the Theoretical Prices of Houses Using the Hedonic Function . . . . . . . . . . . . . . . . . . . . . 2.4.2 Removal of Transactor Error . . . . . . . . . . . . . . . . . . . . 2.4.3 Calculating Gross Returns Utilizing Relevant Information . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .

. .

1 2

. . .

3 4 5

. . . . .

6 7 7 7 8

. . . .

10 13 14 15

. . . . .

17 17 19 23 24

. .

25 26

.

27

xiii

xiv

Contents

2.5 Conclusion . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . Appendix . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . References . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 3

4

Appropriate Housing Valuation Using Hedonic Price Function and Promoting Information Disclosure . . . . . . . . . . . . . . . . . . . . . . . 3.1 Introduction . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 3.2 Housing Valuation Method Using the Hedonic Price Function . . . 3.3 Theoretical Value of Housing Price . . . . . . . . . . . . . . . . . . . . . . 3.4 Appropriate Housing Valuation Method . . . . . . . . . . . . . . . . . . . 3.4.1 Estimation of the Regression Equation, γ, and the Real Theoretical Prices of Houses in the Dataset . . . . . . . . . . . 3.5 Disclosure of Information . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 3.5.1 Information to Disclose and Effects of Information Disclosure . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 3.5.2 Application of the Method to Realize the Real Theoretical Prices of Houses . . . . . . . . . . . . . . . . . . . . . 3.6 Empirical Estimates . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 3.6.1 Estimation of the Theoretical Prices of Houses . . . . . . . . 3.6.2 Estimation of γ . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 3.6.3 Example of the Calculation of the Real Theoretical Prices of Houses . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 3.7 Conclusion . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . References . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . Housing Loan Policy in Japan: Benefits and Drawbacks of Low-Interest Rate Housing Loans by Government Agency . . . . . 4.1 Introduction . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 4.2 Overview of the Japanese Housing Loan System . . . . . . . . . . . 4.2.1 The Japanese Housing Loan System . . . . . . . . . . . . . . . 4.2.2 Public Housing Loan . . . . . . . . . . . . . . . . . . . . . . . . . . 4.3 The Benefits of Direct Housing Loans Issued by the GHLC . . . 4.3.1 Method of Analysis and Preliminary Tests . . . . . . . . . . 4.3.2 Policy Evaluations . . . . . . . . . . . . . . . . . . . . . . . . . . . . 4.4 The Drawbacks of Direct Finance Under the GHLC . . . . . . . . . 4.4.1 Interest Rate . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 4.4.2 The Oppression of Commercial Activities . . . . . . . . . . . 4.4.3 Welfare Loss . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 4.4.4 The General Account’s Compensation for the Deficit . . . 4.5 Conclusion . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . Appendix 1: Data Used for Estimation . . . . . . . . . . . . . . . . . . . . . . . . Appendix 2: Seasonality of Variables in the VAR Models . . . . . . . . . . Appendix 3: Housing Market When the Actual Value of Housing Loans of the GHLC Exceeded the Budget . . . . . . . . . . . . . . . . . . . . . References . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .

29 29 29 31 31 33 36 41 41 41 42 42 43 43 45 45 46 47

. . . . . . . . . . . . . . . .

49 49 52 52 53 55 55 61 69 69 70 70 71 71 72 72

. .

74 74

Contents

5

Implementation of Subsidy for Improving the Earthquake-Proof Conversion of Rental Housing . . . . . . . . . . . . . . . . . . . . . . . . . . . . 5.1 Introduction . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 5.2 Earthquakes in Japan . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 5.3 Method of Analysis and Data . . . . . . . . . . . . . . . . . . . . . . . . . 5.3.1 Discounted Cash Flow Method . . . . . . . . . . . . . . . . . . . 5.3.2 Contingent Valuation Method . . . . . . . . . . . . . . . . . . . 5.3.3 Procedure of Estimation . . . . . . . . . . . . . . . . . . . . . . . . 5.3.4 Questionnaire Survey on the Earthquake-Proof Conversion of Rental Apartments . . . . . . . . . . . . . . . . . 5.3.5 Data . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 5.4 Outcome of Analysis . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 5.4.1 Awareness of the Earthquake-Proof Conversion of Housing . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 5.4.2 Profitability of Earthquake-Proof Conversion of Rental Houses . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 5.5 Factors Influencing the Willingness to Pay for Earthquake-Proof Conversion . . . . . . . . . . . . . . . . . . . . . . 5.6 Subsidy to Accelerate Earthquake-Proof Conversion of Rental Apartments . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 5.7 Conclusion . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . References . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .

xv

. . . . . . .

77 78 81 82 82 82 83

. . .

84 89 89

.

89

.

90

.

91

. . .

94 96 96

. . . . .

99 99 101 103 104

. . . . . . .

104 106 107 108 108 113 113

Correction to: Economic Analysis of Housing Policy in Japan . . . . . . . .

C1

6

Housing Policy to Supply Barrier-Free Rental Housing . . . . . . . . . 6.1 Introduction . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 6.2 Overview of Population Aging in Japan . . . . . . . . . . . . . . . . . . 6.3 The Need to Supply Housing Suitable for the Elderly . . . . . . . . 6.4 The Current State of Barrier-Free Housing . . . . . . . . . . . . . . . . 6.5 Factors That Discourage the Supply of Barrier-Free Rental Housing . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 6.6 Housing Policy Plan to Increase Barrier-Free Rental Housing . . 6.6.1 Method of Funding and Property Management . . . . . . . 6.6.2 Rent Subsidy . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 6.7 Effects of the Policy . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 6.8 Conclusion . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . References . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .

Chapter 1

Estimation and Applications of the Hedonic Function of Housing

Abstract There are several limitations of the conventional housing evaluation methods comprising the cost, sales comparison, and profit return methods. The cost method considers only the cost—the supplier side—and hence it is difficult for buyers to calculate the accrued price. The sales comparison method produces an arbitrary evaluation when attributes of properties to be appraised and comparable properties are different. Additionally, there are cases when comparable or recently sold properties in the area with similar characteristics do not exist. The profit return method does not facilitate an accurate estimation of the future profits moving according to the demand and supply in the rental market. The repeat-sales method suffers from statistical bias because it is restricted to data of houses traded more than twice in a sample period. To overcome these limitations, this chapter proposes the use of hedonic housing valuation. It discusses the definition of the hedonic approach, the method of estimation using the hedonic approach, the merits of using the hedonic housing price and rent function, and the application of hedonic housing valuation and price index. Based on regression analysis, this chapter provides the way to estimate the hedonic function using the housing price and attributes of houses as explained and explanatory variables, respectively. It also regresses the hedonic function with rent and attributes of houses as explained and explanatory variables, respectively. By substituting the level of each attribute in the estimated hedonic price and rent functions, the chapter presents the price and the rent based on the level of each attribute. The results show that the quality-adjusted hedonic price and rent indices capture the transition of housing price and the movement of rents. This hedonic price valuation provides an accurate valuation of used houses, and consequentially contributes to revitalizing the used housing market. Additionally, the broad area is shown in which hedonic housing valuation and hedonic housing price index are applicable. Keywords Hedonic price function · Hedonic method · Housing valuation · Price index · Used house market

© Springer Nature Singapore Pte Ltd. 2022, corrected publication 2022 K. N. Hirono, Economic Analysis of Housing Policy in Japan, New Frontiers in Regional Science: Asian Perspectives 64, https://doi.org/10.1007/978-981-19-4925-8_1

1

2

1.1

1 Estimation and Applications of the Hedonic Function of Housing

Introduction

The hedonic method has several applications in real estate. It can be used for housing valuation and constructing housing price and rent indices. There are several practical uses of the hedonic method such as in natural disaster risk assessment. The results of this analysis using the hedonic method can guide local governments and developers in their efforts to improve living environments of inhabitants. The outcomes of this research can be effective for decision-making in housing trade and investment. Despite its applications, to the best of my knowledge, there are few comprehensive accounts of the hedonic method. For example, Hidano (1997) focuses on the relationship between the hedonic approach and the capitalization hypothesis, and handles cases of valuation of environment and social capital by estimation of the hedonic function of land prices. Kariya et al. (2016) estimate the hedonic rent function in their study on the rental housing market. In the given context, this chapter discusses the following aspects of the hedonic housing function. Firstly, it shows the research process and explained the outcomes. Secondly, it defines the hedonic method. Thirdly, it explains the method to estimate the hedonic housing price function and hedonic price index. Fourthly, it introduces several studies, including those in Japan, using the hedonic method. Fifthly, the chapter discusses the application of the hedonic housing valuation and price index from a broad viewpoint and proposes the use of hedonic housing valuation to vitalize the used housing market. This chapter is a trial to accelerate research on the use of hedonic housing price function and housing price index that have advantages. For example, Bourassa et al. (2016) point out that, although the hedonic price index is used to grasp the fluctuation in the housing price in the United Kingdom, France, Norway, and Switzerland, the repeat-sales method (which uses only samples traded over two times in a sample period) is predominantly used in the United States, where many studies in the housing market are carried out. In this context, the future research can be developed using the hedonic price function and index. The method of estimating the hedonic housing price function and the interpretation of the estimated outcome shown in this chapter provide a basis for Chaps. 2 and 3. The rest of the chapter is organized as follows. Section 1.2 introduces the process of my research on housing valuation using the hedonic function. Section 1.3 reviews related works and the contribution of this study to the literature. Section 1.4 defines the hedonic method. Section 1.5 presents a method to estimate the housing price and the value of attributes using the hedonic method. Section 1.6 presents the hedonic price index. Section 1.7 describes the goods to which the hedonic price index is most applicable. Section 1.8 explains the method for estimating the hedonic price index. Section 1.9 illustrates the examples of the hedonic price index. Section 1.10 proposes the uses and merits of the hedonic housing price and rent function. Section 1.11 shows the applications of the hedonic price index. Section 1.12 concludes the study.

1.2 Process of Research on Housing Valuation Using the Hedonic Function

1.2

3

Process of Research on Housing Valuation Using the Hedonic Function

Ito and Hirono (1993) extend the literature by proposing a fundamental framework and a method of analysis. It has the following components. First, this method calculates the housing price and rent indices, by estimating the hedonic price function. Second, it provides a method to assess the housing price and rent from the attributes and the estimated hedonic housing price and rent functions, respectively. Third, it enables assessing the effect of attributes on housing price and rent by providing an estimation of the hedonic function. The above methods of analysis using the hedonic method are used in the successive chapters in this book. Ito and Hirono (1993) estimate the hedonic function of housing price and rent based on a micro dataset comprising housing price, rent, attributes (e.g., floor space and commuting time, building age, and a dummy variable showing whether a condominium is on the first floor). Hedonic function is the relationship between the price and attributes of goods (and services). Let the goods be houses. Hedonic function is expressed as follows: Pt ¼ hðX 1t , X 2t , . . . , X nt Þ þ U i ,

ð1:1Þ

where Pt is the housing price at time t; X1t, X2t, . . ., Xnt are the attributes of houses (e.g., floor space, building age, and commuting time); and Ui is an error term. In this research, Eq. (1.1) is the fundamental equation. Based on regression analysis, our research estimates the hedonic function using the housing price and houses’ attributes as explained and explanatory variables, respectively. It also regresses the hedonic function using rent and houses’ attributes as explained and explanatory variables, respectively. By substituting the level of each attribute in the estimated hedonic price and rent functions, the research calculates the price and rent of a house in relation to the level of attributes. To the best of my knowledge, our research is the first attempt in Japan to present a method using a hedonic function to assess the price and rent of a house with certain attributes. Second, it devices a method to calculate the effect of changes in the attributes on the price and rent. Third, Ito and Hirono (1993) is the first research in Japan that uses the hedonic method to construct the housing price and rent indices. Fourth, it calculates the return on housing holdings using a micro dataset and rejects the test of weak form efficiency.

4

1.3

1 Estimation and Applications of the Hedonic Function of Housing

Literature Survey

Case and Shiller (1989, 1990) calculate the rate of return on single-family homes and test the weak form of the efficient market hypothesis for Atlanta, Chicago, Dallas, and San Francisco. They use a micro dataset and construct a housing price index based on the repeat-sales method of Bailey et al. (1963). This method involves selecting the repeatedly sold houses and regressing the intertemporal difference in the logarithm of price on the difference in time dummy variables. Since the estimation using the repeatedly sold houses eliminates the need for information about housing attributes, the repeat-sales method facilitates easier estimation than that of the hedonic method. However, Karato (2014) indicates that the use of repeatedly sold houses may lead to statistical bias. In this regard, Karato (2016) shows that the same difference in the building age of two houses and the trade interval leads to complete multicollinearity and the effect of building age on housing price is not identifiable. Linneman (1986) estimates the hedonic functions of housing price, using the data of a survey of owner-assessed housing values in Philadelphia in 1975 and 1978. The semi-strong form of the efficient market hypothesis is tested and not rejected. Concerning rentals, Gillingham (1975) compares hedonic rent functions of 10 US cities—Chicago, Los Angeles, Detroit, Boston, Pittsburgh, Washington, Cleveland, Baltimore, San Francisco, and St. Louis. To the best of my knowledge, this is the first study in Japan that uses the hedonic analysis to estimate the housing price and rent. This literature survey also focuses on various methods to estimate the hedonic housing price function in Japan. For example, Ono et al. (2003) propose the overlapping-period hedonic model (OPHM) that uses pooled cross-section and time series data, and sets a specific period as the estimation period. This estimation involves successive changes in the estimation period and connecting the outcomes of each estimation period. This method yields an index reflecting the structural changes of each period and possessing the merit of smoothness of the index. However, the OPHM lacks theoretical foundations based on macroeconomic and econometrics. For example, the OPHM does not clarify the unbiasedness of the estimated regression coefficient. Maruyama (2004) presents a hedonic analysis employing the Bayesian ridge regression and constructs the hedonic price index of used condominiums. Relative to the ordinary least squares, the ridge regression method solves the problem of multicollinearity. However, the improved ridge regression method suffers from the limitation of functional form; Maruyama (2004) also does not clarify why this method is considered an optimal for hedonic analysis. As stated in the previous section, Ito and Hirono (1993) present a framework using the linear hedonic function to estimate the hedonic housing price and rent indices. This framework has been guiding other related studies such as Nakamura (1996), Tanabe (1994), Suzuki (1995), Nakagami (1995), Omori (2002), Ono et al. (2003), and Maruyama (2004). This method of constructing the price index has been

1.4 The Hedonic Method

5

adopted by research on land price market (Saita, 2004). It has contributed to the development of the corporate goods price index of the Bank of Japan. Because the hedonic method using a linear hedonic function is more tractable and easier to understand compared to OPHM and the method employing the Bayesian ridge regression, the method has been used for constructing the Japan residential property index (JRRPI) by the Ministry of Land, Infrastructure, the Transport and Tourism (MLIT) and the recruit residential price index (RRPI) by the Recruit company. This research discusses the application of Ito and Hirono’s (1993) framework in the literature. Firstly, Ito and Hirono (1993) use the hedonic function to construct a method to estimate the level of price and rent of a house with certain attributes. This method has been used without modification in subsequent studies on the housing market such as those by Nakamura (1998), Yamazaki (1999), Tanabe (1994), Tiwari and Hasegawa (2000), Nakagami (1995), Yamaga et al. (2002), and Nishimura et al. (2002). Secondly, Ito and Hirono (1993) demonstrate how to find the effect of each attribute on housing price or rent, by estimating the hedonic price function or rent function. This method has been used in succeeding studies such as Arai (2006), Saito (2005), and Ootake and Yamaga (2001). Arai (2006) shows a relationship between the rent and the floor space, Saito (2005) analyzes the effect of flooded area on the price of condominiums using a dummy variable showing whether a condominium is built in a flooded area. Ootake and Yamaga (2001) show that the rent of fixed-term rental housing is statistically significantly lower than that of the general rental housing.

1.4

The Hedonic Method

The hedonic method captures the price factors of goods based on their performance or function-related characteristics. It must be understood that consumer utility (satisfaction level) is determined not by the goods themselves but by their performance and functions that are referred to as attributes. Therefore, in the hedonic method, the price of goods is explained by their attributes. This price of goods is valued based on the sum of the attribute values, and the price of each attribute is estimated using regression analysis. In this context, it must be noted that the hedonic method serves as an effective tool for analyzing consumer goods with distinct attributes. For example, an estimation of automobiles’ price would involve a focus on their attributes such as horsepower, model name, and operations (e.g., hybrid or regular vehicle). Similarly, an estimation of price of a personal computer (PC) would entail an assessment of its attributes, including memory, the central processing unit (CPU), the hard disk space, and the size of the liquid crystal display (Shiratuka, 1994). This research focuses on houses. Specifically, Chap. 2 focuses on the main attributes of condominiums in Tokyo, which are building age, floor space, and the time to commute to the center of Tokyo. The consumers in Tokyo attach importance to building age, floor space, and

6

1 Estimation and Applications of the Hedonic Function of Housing

time to commute when they buy a condominium. In the case of a detached house, attributes include living environment factors, in addition to the above attributes. The relationship between the price of a Tokyo condominium and its attributes is expressed as follows: Price of a condominium in Tokyo ¼ f (floor space, building age, time to commute to the center of Tokyo). If the number of attributes is n, then the price of goods will be formulated as the function of the attributes (hedonic price function), as shown in (1.2) in general: Pg t ¼ hðX 1gt , X 2gt , . . . , X ngt Þ þ U i ,

ð1:2Þ

where Pgt is the price of goods at time t; X1gt, X2gt, . . ., Xngt are the attributes of the goods; and Ui is an error term.

1.5

Method for Estimating the Housing Price and Marginal Valuation of Attributes

Let the good be a house, then Eq. (1.2) becomes Eq. (1.1). If the hedonic price function is linear, Eq. (1.1) becomes Pt ¼ α0 þ α1 X 1t þ α2 X 2t þ ・・・ þ αn Xnt þ Ui ,

ð1:3Þ

Equation (1.3) is estimated by the regression method using a micro dataset. The estimated coefficients α0, α1, . . ., αn show the extent to which housing price Pt will change in response to an increase in each attribute. They denote the marginal valuation of each attribute. For instance, assume that n ¼ 3, and the attributes of the houses are floor space, building age, and the time to commute to the center of Tokyo. Then, the estimated coefficients of α0, α1 and α3—the marginal value of floor space, building age, and commuting time—are an increment in the housing price with a one-unit increase in each of these attributes, respectively. In the hedonic method, Eq. (1.3) is used to value the goods as a total of each attribute. By estimating Eq. (1.3) by the regression method, housing price can be estimated as a value of the aggregate of each attribute. The price of a house with certain attributes can be calculated by substituting the level of these attributes X1t, X2t, . . ., Xnt in the estimated equation. If we regress the Eq. (1.4) (in semi-logarithmic form), then the estimated coefficients γ 0, γ 1, . . ., γ n will reveal the percentage of increase in the housing price with a one-unit increase in each attribute. log Pt ¼ γ 0 þ γ 1 X 1t þ γ 2 X 2t þ ・・・ þ γ n Xnt þ Uit:

ð1:4Þ

1.8 Estimation of Hedonic Price Index

7

In Eqs. (1.2)–(1.4), I set the qualitative variables among attributes—e.g., whether a house is south facing—as dummy variables. Chapters 2 and 3 present the outcome of the estimation of the hedonic housing price function using the data of houses along the Yamate, Chuo, Toyoko, Sobu, Keihin Tohoku, and Joban lines in the Tokyo metropolitan area. Let the left-hand side of Eq. (1.3) be rent. This can allow for the estimation of the hedonic rent function and the calculation of rent, given certain attributes and changes in rent owing to an increase in each attribute.

1.6

The Hedonic Price Index

The hedonic price index is an index without variation in prices due to changes in attributes. For example, an increase in the price of PC may practically mean a decline in its price, if there is an improvement in its calculation speed and memory. The rate of the quality-adjusted price change is the observed rate at which the prices change minus the rate of price change due to changes in quality. The studies on the hedonic price index of automobiles and computers are Griliches (1961) and Berndt and Griliches (1990).

1.7

Examples of Goods

Chapters 2 and 3 focus on attributes of houses in Tokyo, which include floor space, building age, and the time to commute to the center of Tokyo in the hedonic price function. The hedonic price index of houses can be calculated by keeping the level of attributes constant. It must be noted that the hedonic method is used to estimate the prices of goods with distinct attributes. Since it is difficult to decompose the attributes of paintings and books we cannot construct the hedonic price index of these goods. Shiratuka (1994) argues that the hedonic price index is most useful for goods with changing attributes such as PC. In this case, the level of attributes should be kept constant in order to facilitate an accurate estimation of the price index. Given this, the attributes of the following goods are kept constant in the Corporate Goods Price Index in Japan: PC, copying machine, laser printer, ink-jet printer, and video and digital cameras.

1.8

Estimation of Hedonic Price Index

To estimate the hedonic price index, I regress Eq. (1.5) using the micro dataset between different time periods. Let the goods be houses.

8

1 Estimation and Applications of the Hedonic Function of Housing

Pt ¼ α0 þ α1 X 1t þ α2 X 2t þ ・・・ þ αn Xnt þ β1 D1 þ β2 D2 þ . . . þ βm Dm þ Uit ,

ð1:5Þ

The estimated coefficients of each attribute α0, α1, . . ., αn would denote an increase in housing price if the level of each attribute increases by one unit. In Eq. (1.5), Dj ( j ¼ 1, 2, . . ., m) represent dummy variables showing houses at time j, and βj  100 % ( j ¼ 1, 2, . . ., m) show the rate of change in the quality-adjusted price at time t, relative to the base point. For example, if the base point of the year is 2000, and D1 is a yearly dummy of 2020, then β1 shows the rate of change in the qualityadjusted housing price. By calculating the rate of change in the quality-adjusted housing price at each time period and by aggregating the outcomes, I estimate the hedonic housing price index. For example, to estimate the annual index for the sample period 2014–2021, it would be critical to include the yearly dummies of 2015–2021, without the yearly dummy of the base point (i.e., 2014), in Eq. (1.5). In this case, the Eq. (1.5) can be expressed as follows: Pt ¼ α0 þ α1 X 1t þ α2 X 2t þ ・・・ þ αn Xnt þ β1 D2015 þ β2 D2016 þ . . . þ β7 D2021 þ Uit , ð1:6Þ I regress Eq. (1.6). The estimated coefficient β1 shows the rate of change in the quality-adjusted housing price in 2015 relative to 2014, and the estimated coefficient β2 shows the rate of change in the quality-adjusted housing price in 2016 relative to 2014, and so forth. β2 – β1 is the rate of change in quality-adjusted housing price from 2015 to 2016. This approach can be used to calculate the rate of change in the quality-adjusted housing price, for each year; the outcome for each year can be aggregated to construct the hedonic housing price index between 2014 and 2021. It must be noted that quarterly or monthly data can be used to calculate quarterly or monthly quality-adjusted price indexes, respectively. The hedonic price index varies according to changes in the factors influencing both the housing demand (e.g., population and income, the interest rate on the housing loan, and the fixed property tax rate) and the housing supply (e.g., the prices of construction materials, and wage costs). Let the left-hand side of Eq. (1.5) be rent. The aforementioned approach can be used to estimate the hedonic rent function and construct a quality-adjusted rent index.

1.9

Examples of Hedonic Housing Price Index

This section focuses on two hedonic housing price indexes—the JRRPI and RRPI. The JRPPI was made public in April 2008 by MLIT (Fig. 1.1). The merits of this index are as follows. First, it provides a broad coverage of residential properties such as detached houses, condominiums, and the housing land in Japan. Second, it is a

1.9 Examples of Hedonic Housing Price Index

9

Fig. 1.1 Japan Residential Property Price Index. Source: Ministry of Land, Infrastructure, Transport and Tourism (2021)

monthly index. Third, the data of the JRRPI is trade prices, not offer prices. The JRRPI also publishes the housing price indexes of regions (e.g., Tohoku, Kanto, Chubu, Kinki Chugoku, and Shikoku), some urban areas, and three prefectures (i.e., Tokyo, Aichi, and Osaka). The limitations of the JRPPI are as follows. First, Takehashi (2013) points out the long time lag (4 months) between housing trade date and the publication of the JRPPI. This can be attributed to the delay in receiving real estate transactors’ survey responses used to construct trading data, including the trade price and attributes. Second, the main index of the JRPPI focuses on the whole of Japan. Given this broad coverage of the JRPPI, the effect of attributes on the prices of houses differs in each regional housing market, which is not reflected in the JRPPI. Third, this index does not regard the living environment factors as attributes of the detached houses. Concerning the RRPI published by the Recruit company, this monthly index covers condominiums in the Tokyo metropolitan area (Tokyo, Kanagawa, Saitama, and Chiba). The merits of this index are as follows. First, there is a shorter time lag between housing data and its publication of the RRPI. This can be attributed to the fact that the RRPI collects data from housing information magazines and the Recruit company’s website. The limitation of the RRPI is that its price data are the offer prices which may be different from the trade prices. However, there is a devise that the unsold houses are deleted from the samples. Second, the samples in the RRPI are restricted to condominiums in the Tokyo metropolitan area, which increases the accuracy of the hedonic price index. The Recruit also publishes the hedonic rent index.

10

1.10

1 Estimation and Applications of the Hedonic Function of Housing

Advantages and Applications of Hedonic Housing Price and Rent Function

This section discusses advantages and applications of the hedonic housing price and rent function: 1. The hedonic price function can be used to value houses. Specifically, the housing price can be calculated by estimating Eq. (1.3) and substituting the level of attribute of the house in the estimated equation. 2. It is able to calculate the appropriate level of rent of a house with certain attributes. This approach can provide clarity to homeowners looking to rent their property. An accurate estimation of the flow of rental income in the future can help lenders to plan their consumption and asset management. 3. When the level of rent and hedonic housing price function, price and hedonic rent function, or hedonic price and rent functions are given, the hedonic method can be used to calculate the price earnings ratio (PER) of a house with certain attributes. It denotes the price (annual) rental ratio showing the profitability of the investment on the house. The PER shows the number of years needed to collect the invested money, if rent is annual rent. For example, an estimation of the PER of a condominium or one room in a rental building in central Tokyo can provide benchmarks for investment decisions in the housing market. 4. The conventional housing valuation methods—the cost, sales comparison, and profit return method—suffer from several limitations. The cost method considers only the cost—the suppliers’ side—and hence it is difficult to calculate the price accrued to buyers. The sales comparison method may produce an arbitrary valuation when attributes of properties to be appraised and comparable properties are different. Moreover, there are cases when comparable or recently sold properties in the area with similar attributes do not exist. The profit return method does not facilitate an accurate estimation of the future profits, which move according to the demand and supply in the rental market. However, the hedonic method allows for a reliable estimation of goods, in the sense that it is based on the microeconomic theory of Rosen (1974) and Lancaster (1971). According to Rosen (1974), the housing price valuation using the hedonic method represents the market clearing price. In other words, this housing price presents an objective valuation in the housing market comprising the houses’ consumers and producers. 5. The hedonic price function captures changes in housing price according to the variation in the attributes of these houses. The estimated coefficients α0, α1, . . ., αn show the effects of the increase in the level of attributes on Pt in Eq. (1.3). In other words, the regression analysis estimating Eq. (1.3) provides the price of each attribute in the housing market. This approach helps Harano (2009) to reveal that the housing prices decrease 1.9% with a one-year increase in the building age of condominiums in the Tokyo wards’ area.

1.10

Advantages and Applications of Hedonic Housing Price and Rent Function

11

6. The hedonic price function can identify attributes influencing housing prices. As for an estimation of Eq. (1.3), including some attributes, only statistically significant attributes influence housing prices. In this regard, it must be noted that the consumers attach significance to certain attributes, which also influence the housing prices. 7. The hedonic price function can capture the value of living environments, which are considered attributes of detached houses and the land for detached houses. Therefore, when estimating the hedonic price function of detached houses or housing land, the living environment factors should be treated as explanatory variables. This approach can figure a differentiation between the housing price of detached houses and the land prices of detached prices, depending on the level of living environment factors. This method is rooted in the capitalization hypothesis, that is, increasing the living environment factors can increase the demand of detached houses or housing land, and thereby increase their prices. Let the price of the housing land (or detached house) be the left-hand side variable of Eq. (1.3). Let us also include the variables of attributes representing the living environments and regress Eq. (1.3). The estimated coefficients of these variables would indicate an increase in the housing land price (or the price of the detached house) in response to a one-unit increase in the variables of living environments. Several studies show the effects of living environments on the price of the housing land or detached house; the results of these studies are summarized as follows. First, Gao and Asami (2005) analyze detached residential blocks in Tokyo using a hedonic pricing model and list the following local attributes influencing the price of residential land: hours of sunlight (over 4 h), parking space, front road width, and neighborhood tree planting. Gao and Asami (2005) conclude that the sub-division of land lots decreases the value of detached residential land. Second, Hasegawa et al. (2007) test the effects of regulation of land use on the price of detached house. They show that the following factors significantly affect the price of detached houses: volume ratio, build rate, and a dummy variable indicating the category 1 low-rise exclusive residential districts and site area restrictions. Third, Hidano (1987) empirically evaluates living environments by estimating the hedonic price function of land price. The study shows that an increase in land price is influenced by accessibility to moderate-scale parks, visibility of parks, and land readjustment. Fourth, Okazaki and Matsuura (2000) study urban planning regulations. They show that the land price is positively influenced by category 1 low-rise exclusive residential districts—indicators of comfortable living space—and volume ratio (which raises the efficiency of land use). Several studies also focus on the effects of living environments on rent when estimating the hedonic rent function. The results of these studies are as follows. First, Kutsuzawa et al. (2007) point out that the crime rates of intrusion theft negatively impact the rent of small-size condominiums on the first floor. Second, Kuroda (2020) presents that the quality of public high school in the school district—the average school-level test score—significantly increases rent.

12

1 Estimation and Applications of the Hedonic Function of Housing (10 thousand houses) 1000

(%)

Number of vacant houses

900

9.4

700

659

10

448 6

330 268

300

4

172 103

52

12

8 394

4.0 2.5

846

576

5.5

400

0

9.8

7.6

500

820 757

8.6

600

100

12.2 11.5

Rate of vacant houses

800

200

13.6 13.1 13.5 14

2 0

Fig. 1.2 Number of vacant houses and the rate of vacant houses in Japan. Source: Ministry of Internal Affairs and Communications (2018)

8. This analysis also focuses on studies considering the living environment factors using the hedonic method. This analysis can guide the decisions of local governments and developers. While it can help the former develop policies to improve the living environments, and thereby their valuation by the inhabitants, it can help the latter to determine the investment that can improve the living environments. For example, Yasuda and Hirono (2016) quantitatively calculate the value of living environmental amenities of the residential lots in the Tokyo metropolitan area, by estimating the hedonic price function. We present several strategies to develop the residential lots. Specifically, we suggest developers to make quiet residential lots, establish medical facilities, promote donation and collection of books, and attract private junior high schools. Concerning the policy of the local governments, we show a significant effect of expanding the range of fire prevention and semi-fire prevention zones. 9. This analysis discusses studies considering the natural disaster risk’s impact on housing and land prices, by estimating the hedonic price function. These studies use variables expressing natural disaster as explanatory variables. Ko et al. (2012) show that the risks emerging from the Uemachi active fault line in the eastern part of the Osaka prefecture significantly influenced the land price, after the earthquake driven by the Rokko-Awaji fault line in 1995. 10. The use of the hedonic housing valuation vitalizes the used housing market (Hirono, 2010). According to the Ministry of Internal Affairs and Communications (2018), the proportion of vacant houses account for 13.6% of the total houses in Japan. The number of vacant houses increased by 4.52 million from 1988 to 2018 in Japan (Fig. 1.2).

1.11

Use of Hedonic Price and Rent Indices

13

However, the proportion of used housing sales accounts for only 14.5% of the total sale of houses (MLIT, 2020). This ratio is lower than that in other developed countries. For example, this proportion stands at 85.9%, 81.0%, and 69.8% in the United Kingdom, the United States, and France, respectively. These figures imply that the used housing market is inactive in Japan. The biggest issue facing the used housing market in Japan is that 47% of the consumers in the existing housing market lack of clarity about the prices of used houses (Recruit Housing Institute, 2008). Hence, I propose the hedonic price function for the valuation of used houses without analytical errors by the way presented in Chap. 3 and the government policy that can disclose the estimated hedonic price function equation, ratio of the analytical error to the total error, and Eq. (1.6), in Chap. 3 (Hirono 2010). This proposal can influence the price of the houses in the used housing market in that the house price can represent the market clearing price. Specifically, the hedonic function will make the consumers aware of the appropriate prices of used houses, which will solve the main problem facing Japan’s used house market and, in turn, revitalize the used housing market.

1.11

Use of Hedonic Price and Rent Indices

The hedonic method can be used to construct the quality-adjusted price and rent indices, as shown in Sect. 1.8. In the case of the quality-adjusted price index, the attributes of goods are kept constant and the prices do not change with a change in the level of attributes. The hedonic price and rent indices can be used as follows: 1. Hedonic housing price index is one of the indexes that constitutes the corporate goods price index. 2. It appropriately captures the transition in housing price. In this regard, it must be noted that, according to Rosen (1974), the price of a house with attributes can be calculated and valued by the hedonic price function. This price represents the market clearing price at which the demand price equals the supply price of houses with certain attributes. Given this, the hedonic housing price index also shows a transition in the level of price, reflecting the supply and demand in the housing market. In this context, it must be noted that an awareness of the appropriate movement of house prices, using hedonic housing price index, can help both the buyers and suppliers of housing to trade houses at ease and in conformance to their plans. 3. As stated earlier, it can be used as an economic indicator of the appropriate level of price reflecting supply and demand in the housing market. It should be understood that the price and value (price times the volume) of housing are important as macroeconomic factors. This is because the price and value of housing influence the economy. First, an increase in the value of housing leads to the wealth effect, which increases the consumption of homeowners. Second, an increase in the housing prices owing to an increase in the housing investment

14

1 Estimation and Applications of the Hedonic Function of Housing

implies occurrence of the multiplier effect, which positively influences income of the economy. These factors call for an accurate estimation of the changes in the housing price. The policymakers should also focus on the housing prices. In 1990, the collapse of the bubble in the prices of stocks, land, and houses in Japan contributed to a successive compound recession. In the United States, the housing price declined in 2007, following a surge in 2000–2006. This led to the subprime mortgage problem, which caused the Lehman’s fall and the world-wide financial crisis. Since boom and bust in the housing market can happen in any country, policymakers and economic entities should monitor the housing price, and hedonic price index can facilitate an accurate estimation of the housing price. For example, among the economic indicators, the Halifax House price index, based on the hedonic method, is considered an important economic indicator by the Bank of England. 4. Halifax House Price Index in the United Kingdom is used as underlying finds of derivatives (Shimizu and Watanabe, 2009). A similar measure can be used to plan financial products using the hedonic housing or land price index. 5. The hedonic housing price index is useful for financial planning of households that may own a house as an asset or plan to inherit a house. 6. The hedonic rent index provides useful information to lenders, borrowers, and investors in the rental housing market, by showing the movement of rents. For instance, the hedonic rent index can influence investors’ decision to invest in rental houses.

1.12

Conclusion

There are several loopholes in the existing housing valuation methods—the cost, sales comparison method, and profit return methods. Concerning the housing price index, the repeat-sales method has both strengths and weaknesses. This research addresses these gaps by promoting the use of the hedonic method for housing valuation and estimation of the housing price index. Firstly, it explains the way to estimate the hedonic housing price function and hedonic housing index. Secondly, it shows the application of the hedonic housing price function and hedonic price index in a wide area of research, business, and policymaking. The future scope of this research is as follows. Firstly, it will verify differences in the valuation of living environments across regions in Japan. As shown in Chap. 2, in the Tokyo metropolitan area, the price of condominiums is most affected by the commuting time to the center of Tokyo, while it is not affected by the living environment factors. In the regions around Osaka and Kobe, the extent of greening influences the housing price, while lifestyle and the sense of value of inhabitants influence the results of the hedonic price function estimated in residential areas (Tokuda, 2011). Secondly, in the hedonic price function for the Tokyo metropolitan area, the coefficient of the commuting time to Tokyo may decrease after the COVID19 pandemic because some companies will continue remote work and the needs of workers to commute to the center of Tokyo may decline.

References

15

References Arai, H. (2006) “Seifutokei niokewru Mochiie no Kizokuyachin nituite,” Jutakutochikeizai (Quarterly Journal of Housing and Land Economics), 61, 26–34. Bailey, M. J., R. F. Muth and H. O. Norse (1963) “A Regression Method for Real Estate Price Index Construction,” Journal of the American Statistical Association, 58, 933–42. Berndt, E. R., and Z. Griliches (1990) “Price Indices for Microcomputers: An Exploratory Study,” NBER Working Paper No.3378. Bourassa, S. C., E. Cantoni and M. Hoesli (2016) “Robust Hedonic Price Indexes,” International Journal of Housing Market and Analysis, 9(1), 47–65. Case, K. E. and R. J. Shiller (1989) “The Efficiency of the Market for Single-Family Homes,” American Economic Review, 79(1), 125–37. Case, K. E. and R. J. Shiller (1990) “Forecasting Prices and Excess Returns in the Housing Market,” AREUEA Journal, 18(3), 253–73. Gao, X. and Y. Asami (2005) “Mikurona Jyukankyo no Hedonic Bunseki,” Shisanhyoukaseisakugaku (Journal of Property Assessment Policy), 7(2), 30–35. Gillingham, R. (1975) “Place to Place Rent Comparisons,” Annals of Economic and Social Measurement, NBER, 4(1), 153–78. Griliches, Z. (1961) “Hedonic Price Indices for Automobiles,” in Z. Griliches eds. Price Indexes and Quality Change, Harvard University Press. Harano, K. (2009) “Jutakukakaku to Hedonic Ho,” Fudosankantei (Real Estate Appraisal), Practical Study Group on Real Estate Appraisal, 46(3), 34–39. Hasegawa, K., M. Tanishita and C. Shimizu (2007) “Hedonic Bunseki to sono Ooyo nikansuru Kosatsu—Daitoshiken niokeru Takuchibaibaidata o Sozaitoshite,” Tochiseisaku nikansuru Kisotekichousagyoumuhoukokusho, Ministry of Land, Infrastructure, Transport and Tourism, 341–88. Hidano, N. (1987) “Jyukankyoseibi to Chikahendo—Amenity o Hyoukasuru,” Fudosankenkyu, 29 (2),1–10. Hidano, N. (1997) Kankyo to Shyakaisihon no Keizaihyoka, Tokyo: KeisoShobo. Hirono, K. N. (2010) “Housing Policy for Utilization of Existing Housing,” Pacific Economic Review, 15(5), 697–707. Ito, T. and K. N. Hirono (1993) “Efficiency of the Tokyo Housing Market,” Monetary and Economic Studies, 11(1), 1–32, Bank of Japan, July 1993 and NBER working paper No. 4382. Karato, H. (2014) “Repeat Sales Kakakushisu niokeru Selection Bias,” Doshisha Shogaku (Doshisha Business Review), 66(1), 27–47. Karato, H. (2016) “Hedonic Approach o Riyoshita Fudosankakakushisu no Suiteihoho to sono Mondaiten,” Toshijutakugaku (Urban Housing Sciences), 92, 17–20. Kariya, T., Y. Kobayashi and C. Shimizu (2016) Chintai・bunjojutaku o Riyoshita Fudosankakakushisu no Suiteihoho to sono Mondaiten, Tokyo: Progress. Ko T., M. Saito, M. Nakagawa and H. Yamaga (2012) “Social Recognition towards Active Fault Risks and its Impact on Land Pricing: a Case of the Uemachi Active Fault Line in Japan,” Ooyochiikigakukenkyu, 16, 27–41. Kuroda, Y. (2020) “Gakko no Shitu ga Fudosanshijou ni Ataeru Eikyo,” Jutakutochikeizai (Quarterly Journal of Housing and Land Economics), 114, 28–35. Kutsuzawa, R., T. Mizutani, H. Yamaga and F. Otake (2007) “Hanzai to Chika・Yachin,” Jutakutochikeizai (Quarterly Journal of Housing and Land Economics), 66, 12–21. Lancaster, K. (1971) Consumer Demand: A New Approach, New York: Columbia University Press. Linneman, P. (1986) “An Empirical Test of the Efficiency of the Housing Market,” Journal of Urban Economics, 20, 140–54. Maruyama Y. (2004) “Ridgekaikisuiteiryo no Riron to sono Ooyo,” Jutakutochikeizai (Quarterly Journal of Housing and Land Economics), 52, 20–28. Ministry of Internal Affairs and Communications (2018) Outline of the Housing and Land Survey, https://www.stat.go.jp/english/data/jyutaku/pdf/30outline.pdf

16

1 Estimation and Applications of the Hedonic Function of Housing

Ministry of Land, Infrastructure, Transport and Tourism (2020) Kizonjutakushijo no Kasseika nituite, https://www5.cao.go.jp/keizai-shimon/kaigi/special/reform/wg6/20200507/pdf/ shiryou3.pdf Ministry of Land, Infrastructure, Transport and Tourism (2021) Residential Property Index, https:// www.mlit.go.jp/totikensangyo/totikensangyo_tk5_000085.html Nakagami, Y. (1995) “Fudosanshijo niokeru Gendaikachimoderu nitusite,” Jutakutochikeizai (Quarterly Journal of Housing and Land Economics), 16, 20–27. Nakamura, R. (1996) Jutakushijo niokeru Manshionkakakukeisei to Syuekiritu nikansuru Kenkyu, Daiichi Jutaku Kensetsu Kyokai. Nakamura, R. (1998) “Manshionkakakushisuu to Shuekisei,” Jutakutochikeizai (Quarterly Journal of Housing and Land Economics), 27, 16–25. Nishimura, K., T. Asami and C. Shimizu (2002) “Fukanzenjoho ga Motarasu Sonshitu: Tokyo Jutakuryutushijo deno Keisoku,” in K. Nishimura eds. Fudosanshijo no Keizaibunseki, Tokyo: Nihonkeizaisha. Okazaki, Y, and K. Matsuura (2000) “Shakaishihontoushi Kankyoyoin to Chikakansu no Hedonic Approach: Yokohama-shi niokeru Panel Bunseki,” Kaikeikensakenkyu, 22, 47–62. Omori, T. (2002) “Tokyoken Manshion Ryutsu Kakakushisu,” Jutakutochikeizai (Quarterly Journal of Housing and Land Economics), 46, 29–35. Ono, H., H. Takatsuji and C. Shimizu (2003) “Kozohenka o Kouryoshita Hedonickata Jutakukakakushisu no Suitei,” Jutakutochikeizai (Quarterly Journal of Housing and Land Economics), 49, 14–23. Ootake, F. and H. Yamaga (2001) “Teikisyakuchikenseido ga Yachin ni Ataeru Eikyo,” JECR Economic Journal, 42, 10–19. Recruit Housing Institute (2008) Kizonjutakuryutukasseika Project Houkokusho (A report on revitalization project of the used housing market), the Recruit Company. Rosen, S. (1974) “Hedonic Prices and Implicit Markets: Product Differentiation in Pure Competition,” Journal of Polital Economiy, 82(1), January/February, 34–55. Saita, T. (2004) “Keibaifudousan kara Mita Syutokenchika no Doko,” Kin-yu Kenkyu, Bank of Japan, 23, 71–113. Saito, R. (2005) “Syutoken niokeru Shinsuikikensei no Chikato heno Eikyo,” Jutakutochikeizai (Quarterly Journal of Housing and Land Economics), 58. 19–27. Shimizu, C. and T. Watanabe (2009) “Nichibei ni Okeru Jutakukakaku no Hendouyouin,” Financial Review, Ministry of Finance, 95, 30–63. Shiratuka, S. (1994) “Bukkashisuu ni Ataeru Hinshitsuhenka no Eikyo—Hedonic Approach no Tekiyo niyoru Hinshituchoseizumipasokonbukkashisu no Suikei—,” Kin-yu Kenkyu, Bank of Japan, 13(4), 61–95. Suzuki, S. (1995) “Jutakushijo niokeru Kakakukeisei no Bunseki—Tokyo-ken niokeru Kakukeisei no Bunseki—.” Financial Review, Ministry of Finance, 34, 91–111. Takehashi, K. (2013) “Atarashii Jyutakukakakushisuu—sono Tokucho to Kadai,” Kin-yu IT Focus (Financial Informarion Technology Focus), Nomura Research Institute, 12–13. Tanabe, W. (1994) “Manshion no Hedonickakaku to Choukasyuekiritu no Keisoku,” Jutakutochikeizai (Quarterly Journal of Housing and Land Economics), 14, 32–39. Tiwari, P. K. and H. Hasegawa (2000) “Tenure Sentaku to Jutakujuyo no Simulation,” Jutakutochikeizai (Quarterly Journal of Housing and Land Economics), 37, 28–35. Tokuda, M. (2011) “Jyutakuchi no Kakakukeiseiyouin ni Kansuru Ichikousatu—Kansai no Ereahikaku wo Case toshite—,” Kenkyu Nenpo, Shiga University, 17, 55–93. Yamaga, H., M. Nakagawa and M. Saito (2002) “Jishinkikendo to Yachin—Shinsaitaisaku notameno Seisakuteki Implication,” JECR Economic Journal, 46, 1–21. Yamazaki, F. (1999) Tochi to Jutakushijo no Keizaibunseki (An Economic Analysis of Land and Housing Markets in Japan), Tokyo: Tokyo University Press. Yasuda J. and K. Hirono (2016) “Jyukankyo Amenity to Jyutakuchikakaku no Hedonic Approach,” Seisakujohogakkaishi (Journal of Policy Informarion), 9(1), 21–25.

Chapter 2

Imperfect Information and Varying Homeownership Investments: Utilizing Deviation from the Rational Expectations Price

Abstract The determinants of homeownership have been undefined despite such purchases accounting for 82.2% of the wealth of a typical household in Japan. Especially, there is no research on the relationship between information and homeownership investment. Hence, I examine the extent to which differences in information regarding housing attributes generate dissimilarities in homeownership investments. I study the advantages accrued to buyers with complete information and who can pay less than the equilibrium price if sellers undervalue their properties. Due to false expectations concerning housing prices, the sellers tend to earn less than the rational expectation price. Likewise, a reduction in homeownership investment can increase buyers’ consumption or investment in other assets. I develop an empirical model to capture the gains of such buyers, which are estimated to be 12.6–27.6% of the equilibrium price of houses at maximum in the Tokyo metropolitan area during the 1980s. In the model, I incorporate the hedonic price function of houses to formulate the theoretical price of a house. During the housing price hike in Tokyo, when the variability of housing prices was large, the case that the buyers’ gain due to the deviation of the housing prices compared to the rational expectations price was large happened. Thus, the disparity in buyers’ gain was evident in the housing market. Keywords Homeownership investment · Imperfect information · Rational expectations

2.1

Introduction

On average, homeownership investments account for 82.2% of the total wealth of Japanese households (Takayama and Arita, 1996). Nevertheless, an investigation of the determinants affecting homeownership investments remains inconclusive. Econlit for American Economic Association (AEA) members reveals no research on the relationship between information and homeownership investment (i.e., homeownership investment refers to the amount paid to purchase a home).

© Springer Nature Singapore Pte Ltd. 2022, corrected publication 2022 K. N. Hirono, Economic Analysis of Housing Policy in Japan, New Frontiers in Regional Science: Asian Perspectives 64, https://doi.org/10.1007/978-981-19-4925-8_2

17

18

2 Imperfect Information and Varying Homeownership Investments:. . .

This chapter explores the extent to which differences in homeownership investment are attributed to the availability of information to house buyers. I assess the gross return of house buyers possessing information about the dwellings, which led to savings in terms of homeownership investment. Specifically, I studied the housing market of the Tokyo metropolitan area.1 In this chapter, “imperfect information” denotes expectations not formed rationally, and that housing prices do not fully reflect information on housing attributes using a market efficiency hypothesis. In particular, “information on housing attributes” does not indicate information related to concrete attributes, such as area or commuting time, but rather signifies a proper evaluation of a house with these attributes, including macroeconomic circumstances. Accordingly, I construct an empirical model to analyze the effect of buyers’ information on homeownership investment (i.e., I determine the advantage accrued to house buyers who have more information). Using the theory of rational expectations, the advantage to house buyers who can evaluate houses properly at a theoretical price, and those who can pay less than the theoretical price when sellers do not form rational expectations and sell houses at prices below the theoretical prices, is estimated. I also explain the source of our data (Shukan Jutaku Joho), and the railway lines in the Tokyo metropolitan area to which the data apply, for readers unfamiliar with the Japanese housing market. According to our empirical findings, during the 1980s, house buyers in Tokyo could receive a gross return of 12.6–27.6% of the theoretical prices of the houses at maximum if they purchased property by utilizing relevant information. Thereby, they could decrease their homeownership investment and increase their consumption or investment in other assets. In this chapter, I explicate that houses existed of which prices established by suppliers largely deviated from rational expectations, and the variation of that deviation was large in 1988 and 1989. Notably, 1988–1989 was the period when the rate of change in housing prices exhibited both positive and negative signs, and housing prices were quite steep. This phenomenon of deviation and its variation is considered to be the outcome of structural changes in the market. In the next section, I develop our empirical model to verify the relationship between acquired information and homeownership investment. In Sect. 2.3, I clarify the housing data used in this study, and in Sect. 2.4, I apply the test to the data of the Tokyo Metropolitan Area compiled from Shukan Jutaku Joho and explain the outcome of the test. Lastly, Section 2.5 provides the conclusions and the way forward.

1 Data and a part of the analysis in this chapter are from Hirono (2004) and Hirono (2012). For an analysis of the housing market in Japan using the hedonic function, see Ito and Hirono (1993).

2.2 The Model

2.2

19

The Model

This chapter studies the extent to which house buyers can alter their homeownership investments by utilizing information related to the attributes of their dwellings. I investigate the profit a buyer can obtain through buying a house that a seller has undervalued. If a buyer can buy an undervalued house, the amount of income spent on housing can be decreased to the extent of undervaluation. Consequently, the consumption or investment in other assets can be increased. Suppose that each buyer purchases one house for a residence. To ascertain the lowest possible amount paid for the houses by the buyers, I compare the actual housing prices with the theoretical prices (i.e., the price fully reflecting the available information). Specifically, I use a hedonic price model to calculate the theoretical prices of houses and the difference between actual housing prices and the theoretical prices. According to the hedonic price model developed by Griliches (1961, 1971) and Rosen (1974), theoretical prices of houses are a function of house attributes (i.e., quality). Rosen (1974) specified that the hedonic price of a good with certain attributes is the supply-demand equilibrium price of the good possessing such attributes. Thus, the rational expectation price of a house equals to a hedonic price of a house. Therefore, the profit of the buyer who properly evaluates a house at a theoretical price, obtained from buying a house at a relatively cheap price compared to the attributes of the house, which is gain accruing to buyers who have complete information, or the value of home ownership investment with rational expectations can save is the difference between the hedonic price of the house and the actual price of the house. Such trades appear when a seller evaluates a house at a price less than the equilibrium price, which perfectly reflects relevant information. The evaluation of a house by sellers and buyers expands significantly (Linneman, 1986). Possible causes of why a seller does not offer a house at theoretical prices are as follows: 1. Sellers of housing may mistake the position of the supply curve and demand curve of housing because of structural changes in the market. 2. Sellers of used houses are often those owners who had been living there for a long time; therefore, these individuals are non-experts in the evaluation of houses. 3. Since the sellers do not adopt a hedonic method of evaluation, which calculates the value of a house from attributes of the house using a hedonic price function, their evaluation deviates from the theoretical prices calculated via the hedonic method. Besides, the newly built houses are mainly appraised using the cost method. As the cost of advertisement and profit differs among companies that construct new houses, the value of new houses calculated by the cost method also differs. Further, the used houses are evaluated through the comparison method of transaction cases wherein the value of a house differs depending on the transaction cases used in this method. These aspects may result in a range of evaluations of the houses.

20

2 Imperfect Information and Varying Homeownership Investments:. . .

Conversely, the cases wherein buyers do not buy houses at theoretical prices are as follows. There are cases where buyers of houses collect information from housing informational magazines, leaflets, and real estate agents, and buy bottoms. When sellers evaluate and sell houses at prices lower than the theoretical prices, as in the aforementioned cases of a, b, and c, buyers who can evaluate houses properly can utilize this opportunity (e.g., buying bottoms). Applying Rosen (1974)’s model to the housing market, the hedonic price function of houses is log e Pt,i ¼ f ðX 1 , X 2 , . . . , X n , θt Þ

ð2:1Þ

where Pt, i is the market price of a house i at time t; X1, X2, . . ., Xn are attributes of the house (e.g., area, commuting time, and age), and θt is the quality-adjusted price index at time t, which moves according to the shifting of parameters of the housing demand and supply functions. To calculate the difference between the actual and theoretical prices of houses, I first estimate Eq. (2.1), utilizing yearly dummies to estimate the quality-adjusted price index. I regress (2.1) and calculate the theoretical price of a house by substituting X1, X2, . . ., Xn,, and yearly dummies in the estimated equation. Let LUi denote the difference between the actual and theoretical prices of houses. The theoretical price of a house fully reflects information concerning the qualities and attributes of the house. Provided there is no estimation error, LUi ¼ 0 at this theoretical price. Assume Ui ¼ (Pt,ai  Pt, i) /Pt,ai as the difference between the actual price Pt,ai and the theoretical price of a house (transformed by exponential function) Pt, i divided by the actual price.2 If a buyer can obtain a gross return using information regarding the qualities of a house, Ui is negative. If there is an error in the estimation of the hedonic function, I can remove that from the error as a whole. I divide the whole error into analyst error and transactor errors using Linneman (1986)’s method. The analyst error is an estimation error of the hedonic function (e.g., a measurement error, inadequate sampling, or an incorrect form of hedonic function). Contrariwise, a transactor error is caused by the overvaluation or undervaluation of a house. A transactor error arises when a buyer purchases a house at a price higher than the market price, or when a seller sells a house at a price lower than the market price. In summary, the transactor error is induced by traders’ imperfect information. I assume that the transactor error is – α times the whole error (i.e., the sum of the transactor error and the analyst error) Ui (1  α  0). According to Linneman (1986), let – α be constant for all houses. Afterward, a test is performed to check whether α changes according to market conditions. That is, I tested the possibility that α depends on the yearly dummies. The outcome is that α does not depend on the yearly dummies (see the Appendix).

2

Since I use the results in Table 2.2 for our calculation of α, I divide this by the actual prices.

2.2 The Model

21

Pt, i denotes the theoretical price of house i at time t without the whole error, and I call this price the real theoretical price. Here, the ratio of the whole error to the transactor error is:

Pt, a i  Pt  , i : Pt, a i  P t, i ¼ 1 : α Alternatively, α is the ratio of estimation errors of the hedonic function among the whole error. Provided there is no analyst error, α ¼  1, and if there is no transactor error, α ¼ 0. When the transactor error αUi is positive, Pt,ai > Pt, i. If so, the transactor error is an error made by a buyer who purchases an overvalued house compared to the market price. The transactor error αUi is negative when a seller sells a house at a price lower than the market price. I calculate the gross return obtained by the strategy of buying a house when a seller sells it at an undervalued price. Let NRi be the gross return (NRi > 0). Since NRi is the return from a transactor error, NRi can be written as follows:

 a a a NRi ¼ P t, i  Pt, i ¼ ð1Þ Pt, i  Pt, i ¼ aU i  Pt, i ,

ð2:2Þ

where αUi is multiplied by Pt,ai to display the value of NRi in yen. As the return obtained through this strategy is the transactor error in the case of (–α)Ui < 0, the gross return represents the profit from buying a house that is undervalued relative to its quality. I can estimate α from the data since the n-period return of houses changes as much as αUi when the whole error of house i, divided by its actual price, is Ui . Pat + n, i denotes the price of house i at time t + n. The n-period return on houses is defined by:
Pa tþn, i  Pt, a i =Pt, a i :
Return ¼ Pa tþn, i  Pt, a i =Pt, a i  a
 a  ¼ P tþn, i  Pt, a i  P t, i  Pt, i =Pt, i  a  a ¼ P tþn, i þ aU i  Pt, a i  P t, i =Pt, i a ¼ Pa tþn, i =Pt, a i þ aU i  P t, i =Pt, i :

ð2:3Þ

I then calculate α by regressing the return on Ui in (2.3). Since the correlation coefficient of Pat + n, i /Pt,ai and Ui is 0.167 (i.e., is a small value) in the data set in Sect. 2.3, I can assume that these are not correlated in this regression. Namely, there is no reversion in the housing market in the sense that t + n house prices, in proportion to Pt,ai , remain unaffected by Ui. The return here is the so-called appreciation return. Generally, rent is included in the return of a house. Nevertheless, in this chapter, I exclude the analyst error from the whole error by regressing the appreciation return on Ui and thus calculate α.

22

2 Imperfect Information and Varying Homeownership Investments:. . .

Therefore, I do not include rent or mortgage payments in the return of houses. According to the estimation in Sect. 2.4, a ¼  0.527. I express this gross return in proportion to the theoretical price of the house as follows: NRi  ¼ NRi =Pt, i :

ð2:4Þ

Equation (2.4) illustrates how cheaply a customer could buy a house or to what degree was the house price lower than the theoretical price of the house that fully reflected the information related to its qualities. A buyer who could obtain a gross return of (2.4) using such information could decrease the amount needed to spend on housing. Since the purpose of this chapter is to determine the advent of differences in homeownership investments due to varying degrees of information known by the buyer, I calculate the maximum value of NRi. For instance, the actual price of a high-rise condominium along the Toyoko Line was ¥33.8 million in 1988. The attributes of the condominium include a commuting time of 56 min to central Tokyo, floor space of 54.35 m2, and building age of 6.1667 years. I can estimate the theoretical price of this condominium, Pt, i as ¥39.40835 million using the estimated equation of the hedonic function given in Sect. 2.4 (refer to the section on the estimation method). I obtain Pt, a i =Pt, i ¼ ¥5:60835 million and Ui ¼  0.1659. A buyer could acquire a gross return of NRi ¼ ¥2.9556 million, and therefore NRi ¼ 0.075 (i.e., 7.5% of Pt, i ). I analyze the housing market of the 1980s3; there was not much change in house prices during the first half, but the latter half of the decade depicts a bubble in Japan when significant hikes in house prices occurred.4 Bubble or inflated prices were due to great demand and willingness of consumers to pay high prices. This situation was marked by a period of positive fluctuations in house prices. It might have been difficult for sellers to reflect the true market prices of their properties (i.e., prices that fully reflect information) in their pricing schemes since housing prices fluctuated widely. Hence, I analyzed and compared a period of great change in house prices with a period of small change in such prices.

I did not use houses listed in Shukan Jutaku Joho in 1981–82 as data because of incomplete listings. 4 The rate of change in house prices is calculated using the hedonic price function in this area in Hirono (1998). 3

2.3 Data

2.3

23

Data

I compiled data from houses listed for sale in Shukan Jutaku Joho. The estimation period was between 1983 and 93; our data comprised samples taken from the first and second weeks of the calendar year, which are usually sold in the third and fourth weeks of a given year. Shukan Jutaku Joho was a widely circulated weekly magazine during the estimation period, which listed houses for personal use, rent, and investment. It enjoyed wide sales to individual buyers and real estate agents. All bookstores and railway station kiosks sold them. Each listing of a house includes the price, address, floor space, commuting time to the nearest station by bus and/or on foot, a floor plan, construction information, availability of parking space, building age, aspect, and other attributes, such as whether or not it had been renovated recently, and whether it was situated on a corner. Since I calculate returns based on the prices of houses at time t and once again 3 years later, I use data on the same condominiums listed in Shukan Jutaku Joho in a similar time frame. By “the same condominiums,” I mean that the following attributes are the same: address, amount of floor space, balcony area, aspect, floor plan, and construction; differences in building age had to be approximately 3 years.5 All housing samples are from high-rise residential condominium units. That is, dormitories, corporate housing, offices, apartments, and shops are excluded. I use housing units along the Yamate, Chuo, Toyoko, Sobu, Keihin Tohoku, and Joban Lines in the Tokyo metropolitan area. Figure 2.1 shows a schematic map of these railway lines. Commercial districts, such as “old” Tokyo, and many large and small office buildings are located within the 35-km Yamate Loop. Many major firms have headquarters situated in Otemachi and Marunouchi, which are near the Tokyo Station on the Yamate Line. There are terminals of several suburban commuter railway lines, such as Shinjuku and Shibuya, which are major transfer stations from suburban commuter lines to subways or the Yamate Line (Ito and Hirono, 1993). The Chuo Line extends from Shinjuku Station to the west of the Tokyo Station. Thus, it is convenient for residents along the Chuo Line to commute to Tokyo Station. The Toyoko Line stretches from the Shibuya Station in the west to Yokohama in the Kanagawa Prefecture. There are fairly exclusive residential districts along this line. Many commuters from the Chiba Prefecture to Tokyo Station rely on the Sobu Line, and there is a huge commuter belt around this line whereby residents can speedily and conveniently commute from here to Tokyo Station without the transfer. Contrarily, the Keihin Tohoku Line extends to the north from Akabane on the Yamate Line to the Oomiya Station. It takes only 29 min to travel from Akabane to 5

Detached houses are not used as samples since there were no detached houses that were listed in Shukan Jutaku Joho three years later.

2 Imperfect Information and Varying Homeownership Investments:. . .

KEIHIN TOHOKU LINE

24

↑ Oomiya

JOBAN LINE CHUO LINE

Nippori

Shinjuku

←Yokohama

SOBU LINE

Shibuya

TOYOKO LINE

YA M

Tokyo Station AT E

E

LIN

Fig. 2.1 Major railway lines in the Tokyo metropolitan area

Tokyo Station using the Yamate Line, thus making it convenient to commute from the Saitama Prefecture to the center of the business district. The Joban Line stretches northeast from Nippori on the Yamate Line. Typically, rents and house prices are low along the Joban Line compared to the lines on the western side of the Tokyo metropolitan area. The size of the samples is 38 for 1983, 39 for 1984, 51 for 1985, 36 for 1986, 32 for 1987, 68 for 1988, 45 for 1989, and 39 for 1990, 348 in total.

2.4

Empirical Findings

In this section, I calculate the theoretical prices of houses through the hedonic approach in Sect. 2.2 using the data of houses for sale in Shukan Jutaku Joho as explained in Sect. 2.3. First, I calculated the differences between the actual and theoretical prices of houses. Next, I omitted analyst errors from this difference. Then, I exhibited the maximum profits, in proportion to the theoretical prices, accrued to buyers using information about the quality of houses. By utilizing such information, buyers can change their homeownership investments by this amount. Besides, I show the implications of our empirical findings using the theory of rational expectations.

2.4 Empirical Findings

25

Table 2.1 Estimated hedonic function of houses for sale

Condominium at a corner

–

Spec. (2) 7.305 (95.868) 0.013 (12.699) 0.020 (17.084) 0.014 (4.058) 0.529 (13.758) 0.620 (16.443) 0.477 (12.034) 0.095 (1.803) –

Condominium tiled

–

–

R2 SEE

0.837

0.838

0.138 (1.871) 0.078 (0.838) 0.838

0.246

0.245

0.245

Constant Commuting time Floor space Building age Sobu line dummy Joban line dummy Keihin Tohoku line dummy

Spec. (1) 7.309 (95.628) 0.013 (12.731) 0.020 (17.027) 0.013 (3.827) 0.538 (14.058) 0.622 (16.451) 0.479 (12.051)

Recently renovated

Spec. (3) 7.299 (95.483) 0.013 (12.662) 0.020 (16.742) 0.013 (3.625) 0.523 (13.438) 0.609 (15.851) 0.466 (11.559) –

Notes: t-statistics in brackets Recently renovated: a dummy variable ¼ 1 if condominium has been renovated recently Condominium at a corner: a dummy variable ¼ 1 if condominium is at a corner Condominium tiled: a dummy variable ¼ 1 if condominium is tiled

2.4.1

Calculating the Theoretical Prices of Houses Using the Hedonic Function

Using data from 1983 to 1990, I estimated the hedonic function of Eq. (2.1). The estimation results are listed in Table 2.1. The “commuting time” in Table 2.1 refers to the duration in minutes to get to Tokyo, Otemachi, or Hibiya Stations, which is the sum of minutes traveled by foot, bus, and train. Other attributes are floor space (in square meters), building age, and dummy variables, which are set as 1 when a condominium is along the Sobu, Joban, or Keihin Tohoku Lines, respectively (see Table 2.1, Spec. (1)).6 Altogether, two different specifications were used to verify the robustness of the results. Specs. (2) and (3) include dummy variables indicating whether a

6

The Toyoko and Chuo Line dummies are insignificant.

2 Imperfect Information and Varying Homeownership Investments:. . .

26 Table 2.2 Regression of return on Ui

dReturn Results

dUi 0.527*

t-statistics 6.520

R2 0.697

Note: *Statistically different from zero at the 5% level

condominium has been recently renovated, is located on a corner, or is tiled. I include the yearly dummy variables of 1984–1990 in the regression equations of Table 2.1. The regressions are successful, judging from the fact that the adjusted R2s and signs of coefficients concur with our theoretical predictions. According to the results of Spec. (1) in Table 2.1, every additional 10 min of commuting time depreciates the value of condominiums by 13%, an increase in floor space of 10 m2 raises the price by 20%, and condominiums depreciate 1.3% in value each year. Provided the condominium had been recently renovated, the price is 9.5% higher than that of other comparable condominiums (Spec. (2) in Table 2.1): Those located on a corner are valued 13.8% higher in Spec. (3), which may be because they are the sunnier units. Other attributes mentioned in Shukan Jutaku Joho but found insignificant in regressions are the amount of closet space, storage space availability, presence of a small garden, two-sided balcony, area of the balcony, availability of parking space, situated on the first floor or not, construction (i.e., RC, PC, or not), aspect (e.g., facing east, facing south), and near a station where only local trains stop or not. I calculated the theoretical prices of the houses in natural logs by substituting X1, X2, . . ., Xn, and yearly dummies into the estimated equations of Spec. (1) (Table 2.1). Let Ui denote the difference between the actual prices of houses, Pt,ai , and the theoretical prices of houses transformed by an exponential function, Pt, i .

2.4.2

Removal of Transactor Error

I assumed n ¼ 3 and regressed a three-year return on houses (“Return”) on Ui of the hedonic function. Table 2.2 presents the results of the regression. Since there is a large fluctuation in return on houses, I included yearly dummies as explanatory variables in the regression equation in Table 2.2. The table shows that a ¼  0.527, and the coefficient of Ui is significant. I calculated NRi using α, Ui and Eq. (2.2). Table 2.2 shows –1 < α < 0; therefore, the whole error is found to be a sum of the transactor error and the analyst error.7 The reason for this relationship is that the

7 For example, if the empirical findings show a < – 1, the whole error is the difference between the transactor error and the analyst error.

2.4 Empirical Findings Table 2.3 Statistics for gross return utilizing information

27 Year All years 1983 1984 1985 1986 1987 1988 1989 1990

Maximum value 0.276 0.207 0.172 0.219 0.219 0.126 0.276 0.225 0.170

Standard deviation 0.191 0.079 0.066 0.087 0.085 0.107 0.383 0.129 0.081

transactor error is – α times the whole error (0 < – α < 1), and the analyst error is the difference between the whole error and the transactor error.

2.4.3

Calculating Gross Returns Utilizing Relevant Information

In Table 2.3, I provide the summary statistics of NRi, which is the gross return that buyers could obtain by utilizing information on the attributes of houses. I calculated the gross return using Eq. (2.4) in Sect. 2.2. Table 2.3 reveals that buyers could obtain a gross return of 12.6–27.6% at maximum, compared with the theoretical price of houses, by utilizing information related to the attributes of a house. The customers could decrease their homeownership investments by this amount and could use the sum saved to increase their consumption or their investment in other assets. The maximum value of gross return in Table 2.3 shows differences in homeownership investments created by varying information concerning the attributes of houses. Pt, i is the real theoretical price of house i at time t with attributes X1, X2, . . ., Xn without the whole error, and Pt, i corresponds to the intersection of the supply curve and the demand curve of the house with these attributes. According to idea of rational expectations of Muth (1961), when expectations are formed rationally, the predicted prices of suppliers coincide with Pt, i, which is the intersection of the supply curve and the demand curve on average (i.e., Pt, i is the rational expectations price). In the case where a supplier of a house predicts a housing price lower than Pt, i and sets price at this lower level, a buyer with perfect information can raise profits of NRi; that is, the deviation of Pt,ai from Pt, i, which is the supplier’s error regarding expectation. Contrarily, when a house supplier predicts a housing price higher than Pt, i and sets the price at this level, a buyer without information might suffer a loss of NRi ; that is, deviation of Pt,ai from Pt, i (i.e., the supplier’s error as for forecast). In both cases, housing prices deviate from the rational expectations price.

28

2 Imperfect Information and Varying Homeownership Investments:. . .

Thus, profits attained by the buyer using information concerning the attributes of a house is the return yielded from the deviation of housing price from the rational expectations price of the house. The buyer can then decrease the homeownership investment. According to Table 2.3, the maximum value and the standard deviation of the gross return utilizing information were large in 1988 and 1989 compared to the former half of the 1980s when house prices fluctuated slightly. In the late 1980s, which was a period of widely fluctuating house prices, it was difficult for sellers to sufficiently learn the structure of the housing market (i.e., the location of the demand and the supply functions of houses). Additionally, there were structural changes in the market during this period in Japan, which led to sellers’ difficulty in forecasting the location of the demand and supply curve of houses, and sellers’ offering prices deviated from the rational expectations price that provided an opportunity for gross return. In the late 1980s, there were sellers who were offering their houses at higher prices than the theoretical prices and those who were offering them at lower prices in the housing market, which caused a dispersion of sellers’ forecasts of housing prices. The structural changes in the Japanese housing market in the late 1980s are as follows. The Bank of Japan intervened and supported the dollar in the foreign exchange market. As a result, the money supply in Japan rapidly increased, and investment in houses, land, and stocks surged. There were price hikes in stocks, housing, and land, yielding the so-called period of an economic bubble (Miyazaki, 1992). In the late 1980s, the speculative demand for housing increased in the Tokyo housing market. Speculators who bought houses to earn capital gains appeared in the Tokyo housing market, which made housing prices seemingly unrealistic. This situation caused a spiral in the housing market, wherein the speculative demand for housing increased. The period of housing price hikes was from 1986 to 1990. During 1988 and 1989, housing prices were high and generally incremental, but there were quarters when housing prices decreased (Hirono, 1998). A surge in housing prices was evident, but at the same time, housing prices fluctuated as well. In such circumstances, housing prices determined by suppliers deviated from the rational expectations price and scattered, and there existed housing prices that were largely lower than the theoretical price. A housing buyer could get a profit by utilizing this opportunity and decrease homeownership investments. In 1990, when housing price fluctuations stopped, the deviation of housing prices from the rational expectations price decreased, and the gross return utilizing information diminished, as demonstrated in Table 2.3.

References

2.5

29

Conclusion

I applied our model to the housing market in the Tokyo metropolitan area during the 1980s. Differences in homeownership investments arising from buyers’ dissimilar information regarding the quality of houses were 12.6–27.6% of the theoretical prices at maximum. There is also a dispersion in buyers’ gross returns attained by information about house quality. In 1988 and 1989, there was a surge and fluctuation in housing prices in Tokyo, and buyers’ rate of return using information for property varied. Moreover, houses with a large return rate became evident. Interpreting the outcome from the theory of rational expectations, a surprise about structural changes in this period made housing prices deviate largely from the rational expectation price and the deviation scattered. A topic for the future study would be to analyze whether the same phenomenon occurs in the stock market when stock prices are soaring.

Appendix Using the method of including yearly dummies in α in Eq. (2.3), I empirically checked whether α depends on the market condition of each year. The outcome is that there are no statistical differences in α in each year. If α depends on each year’s market condition, α is time-dependent, and the coefficients of yearly dummies should be statistically different from zero in Eq. (2.5). I regressed the return on Ui in Eq. (2.3), including Eq. (2.5) in α of Eq. (2.3) α ¼ α1 þ α2 D84þ α3 D85 þ ⋯ þ α8 D90

ð2:5Þ

where Dmn is the yearly dummy of the year 19 mn. The outcomes are that a2, a3, ⋯, a8 are not statistically different from zero at the 5% level (i.e., t-value is 1.78, 0.97, 1.59, 1.10, 0.51, 0.09, and 0.21, respectively), which means that α does not depend on the yearly dummy of each year.

References Griliches, Z. (1961) “Hedonic Price Indices for Automobiles: An Econometric Analysis of Quality Change,” in The Price Statistics of the Federal Government, General Series, No. 73, NBER, reprinted in Griliches (1971). Griliches, Z. ed. (1971) Price Indexes and Quality Change, Cambridge, MA: Harvard University Press. Hirono, K. N. (1998) “Jutakukakakushisu to Chintairyoshisu no Suikei [Estimation of Housing Price Index and Rent Index],” Josai University Bulletin, Josai University, 16(1), 69–78.

30

2 Imperfect Information and Varying Homeownership Investments:. . .

Hirono, K. N. (2004) “Imperfect Information and the Amount of Housing Ownership,” Pacific Economic Review, 9(4), 335–43. Hirono, K. N. (2012) Jutaku no Shitu ni Kansuru Keizaibunnseki, Tokyo: TagaShuppan. Ito, T. and K. N. Hirono (1993) “Efficiency of the Tokyo Housing Marke,” Monetary and Economic Studies, 11(1), 1–32. Linneman, P. (1986) “An Empirical Test of the Efficiency of the Housing Market,” Journal of Urban Economics, 20(2), 140–54. Miyazaki, G. (1992) Fukugoufukyo [Combined recession], Tokyo: ChuoKoronsha. Muth, J. F. (1961) “Rational Expectations and the Theory of Price Movements,” Econometrica, 29 (3), 315–335. Rosen, S. (1974) “Hedonic Prices and Implicit Markets: Product Differentiation in Pure Competition,” Journal of Political Economy, 82(1), 34–55. Takayama, N. and T. Arita (1996) Chochiku to Shisankeisei: Kakeishisan no Maikurodeta Bunseki, Tokyo: Iwanami Shoten.

Chapter 3

Appropriate Housing Valuation Using Hedonic Price Function and Promoting Information Disclosure

Abstract This chapter proposes a housing valuation method using a hedonic price function obtained from a sample of Tokyo-based condominiums listed for sale in a housing magazine. The chapter calculates the real theoretical prices of houses by regressing the hedonic price function and by substituting the attributes and qualityadjusted price index in the function’s estimated equation, and removing the analyst error in the process. The results show that this valuation method leads to pareto optimal allocation in the housing market. To realize this appropriate housing valuation, information to be disclosed is the estimated equation, parameters used for calculating the real theoretical prices of houses, and the attributes and offer prices of houses. The disclosed information allows for an objective valuation of houses outside a dataset as well as houses inside a dataset. This method can also accelerate the circulation and promotion of used houses and the newly built house market, respectively. Keywords Hedonic price function · Housing valuation · Information disclosure

3.1

Introduction

Housing price valuation is playing an increasingly important role in promoting the resale of used houses and the market for high-quality newly built houses. In this context, it must be noted that information availability and an appropriate valuation method are critical to housing valuation. However, Chap. 1 shows the limitations of the following conventional housing evaluation methods. Firstly, the sales comparison method makes it difficult to find comparable houses. Secondly, the profit return method is not based on an accurate estimation of the expected profits. Thirdly, the cost method focuses only on building firms and not on their consumers (i.e., consumer behavior). There has also been a lack of microeconomic analysis of housing valuation and the disclosure of information. Thus, in the housing policy literature, there is no microeconomic argument about appropriate housing valuation and which information should be disclosed for this valuation. The literature has also failed to provide a © Springer Nature Singapore Pte Ltd. 2022, corrected publication 2022 K. N. Hirono, Economic Analysis of Housing Policy in Japan, New Frontiers in Regional Science: Asian Perspectives 64, https://doi.org/10.1007/978-981-19-4925-8_3

31

32

3 Appropriate Housing Valuation Using Hedonic Price Function and. . .

method to calculate the real theoretical price—which shows the appropriate house value—using a dataset of houses. Chapter 1 proposes a housing valuation method using a hedonic price function. Specifically, it evaluates housing prices based on theoretical prices obtained from the equation regressing housing prices on the houses’ attributes and the quality-adjusted price index (i.e., yearly dummies). In Chap. 3, I remove the analyst error in this procedure. First, this chapter proves that the use of hedonic price function for housing price valuation leads to a pareto optimal allocation in the housing market. In this regard, it extends the microeconomic analysis of Rosen (1974) that the prices evaluated by the hedonic price function represent the market equilibrium between the sellers and buyers. Although the hedonic price function imitates the market price in a perfectly competitive market, the market price does not necessarily lead to a pareto optimal allocation. However, this chapter shows the pareto optimality of this equilibrium. This finding suggests that the valuation method can contribute toward the development of an efficient housing market. Secondly, the proposed method in this chapter requires not only prior estimation of the hedonic price function but also correction of the analyst error. Since this error causes the theoretical prices of houses to diverge from their real theoretical prices in the theoretical model, the error is removed from the theoretical prices calculated by substituting the attributes of houses in the hedonic price function’s estimated equation. I devise a housing valuation method estimating the hedonic price function and removing the error, thereby improving the general housing valuation method based on the hedonic price model. Specifically, this chapter presents the equation to estimate the real theoretical prices of houses (Eq. (3.6)) by removing the transactor error (the error of house sellers and buyers) and the analyst error (error resulting from imperfect information when estimating the hedonic price function). It formulates several equations to estimate the ratio of the analyst error to the total error beforehand. Thus, using Eq. (3.6), housing prices can be estimated accurately. Thirdly, this method facilitates appropriate house valuation (the real theoretical value of houses) based on arbitrary housing data. In this regard, it must be noted that the real theoretical value of a house is its equilibrium or market clearing price. Overall, this chapter regresses the housing price equation on the yearly dummies and house attributes (e.g., commuting time and floor space) and lets the theoretical prices of houses be prices with zero residuals. Fourthly, based on the extension of Rosen’s (1974) model, this chapter shows that a government can achieve pareto optimal housing prices representing an equilibrium between housing demand and supply. Specifically, the government can formulate a policy to disclose the estimated hedonic price function equation, ratio of the analyst error to the total error, and Eq. (3.6), and to promote the use of hedonic housing valuation method. Using this disclosed information, the public sector can calculate the real theoretical price of houses. This policy outcome and the implications of findings in Chap. 2, which showed the difference in the level of housing ownership between buyers with and without information, can contribute toward correcting the inequality in housing ownership caused by uneven information distribution.

3.2 Housing Valuation Method Using the Hedonic Price Function

33

Finally, the chapter shows that the information disclosed for housing valuation includes the estimated hedonic price function equation, parameters used for calculating the real theoretical prices of houses (i.e., equilibrium prices without the transactor and analyst errors) and the information of each house (i.e., the attributes and offer prices), and Eq. (3.6). The rest of the chapter is organized as follows. Section 3.2 develops the housing valuation method, by using the hedonic price function and removing the analyst error. Section 3.3 extends Rosen’s (1974) model by showing the pareto optimality of housing prices that are valued by the hedonic price function. Section 3.4 presents a method to determine the real theoretical prices of houses. Section 3.5 proposes information disclosure and its various components. Section 3.6 applies the method to a dataset. Section 3.7 presents the conclusion.

3.2

Housing Valuation Method Using the Hedonic Price Function

Griliches (1961) and Rosen (1974) developed the basic model of hedonic prices. According to Rosen (1974), the hedonic price function represents the relationship between the prices, attributes, and quality-adjusted price index of various models or various goods and services (with different attributes). The hedonic price function (without error term) is expressed as follows: Pricet ¼ h ðX 1 , X 2 , . . . , X n , νt Þ:

ð3:1Þ

where Pricet is vector of the market price of the model or product variety at time t; X1, X2, . . .Xn are the quality attributes of the goods. For example, if Pricet is the price of a model of PC, the attributes would include memory, processor speed, and the size of the frame. Let νt be an index showing a change in price without a change in qualities—quality-adjusted price index. I express the left-hand side of Eq. (3.1) in a logarithmic form when estimating the equation.1 Applying this theory to the housing market, the hedonic price function of houses is Pt ¼ Pt ðz1 , z2 , . . . , zn , ηt Þ,

ð3:2Þ

where Pt, i is time t market price of a house i; z1, z2, . . ., zn are the attributes of this house (e.g., floor space, commuting time, and building age); ηt is quality-adjusted

1

Data and a part of analysis in this chapter are from Hirono (2004) and Hirono (2012).

34

3 Appropriate Housing Valuation Using Hedonic Price Function and. . .

price index at time t, which moves according to the shift parameters of the demand and supply functions ( αt and βt, respectively, in Sect. 3.3).2 I calculate the theoretical prices of houses, the difference between these prices and the actual prices of houses. First, I regress Eq. (3.2). I use yearly dummies in Eq. (3.2) to estimate the quality-adjusted price index. The estimated coefficient of each attribute of Eq. (3.2) δj ( j ¼ 1, 2, . . ., n) is the rate of change in the housing price with a one-unit increase in zj (the level of jth attributes). I regress Eq. (3.2) and let LUi denote the difference between the actual and theoretical prices of houses, where the theoretical prices are calculated by substituting house i’s attributes—z1 , z2 , . . . , zn and yearly dummies—in the estimated equation.3 If there is no estimation error, these theoretical prices will fully reflect the information on housing quality; at these theoretical prices, LUi ¼ 0. I propose the housing valuation method that values each house at this theoretical price after removing the analyst error. This can be attributed to the fact that, in the absence of the analyst error, these theoretical prices hold optimal values (as shown in the next section).
Let V i ¼ Pat, i  Pt, i =Patq, i be the difference between the actual price of house i at time t  Pat, i and the theoretical price of this house (transformed by exponential function) Pt, i (divided by the actual price of this house at time t  q).4 Next, I use the method of Linneman (1986) to divide the total error into analyst error (caused by errors when estimating the hedonic price function) and transactor error (accompanying the overvaluation or undervaluation of houses). The analyst error refers to the erroneous estimation of the hedonic price function and includes measurement errors, inadequate sampling, and the incorrect hedonic function form. The transactor error occurs when a buyer buys a house at a price higher than the market price or when a seller sells a house at a price lower than the required market price. Briefly, the transactor error is caused by the imperfect information of traders. I assume that the transactor error TRi is γ times the total error (i.e., the sum of the transactor and analyst errors) Vi(0 ≦ γ ≦ 1).5 TR i ¼ γVi ,

ð3:3Þ

When there is no analyst error, γ ¼ 1. Without the transactor error, γ ¼ 0. P t, i denotes the theoretical price of house i at time t in the absence of the total error, and I call this the real theoretical price. Here, the total error: transactor error ¼ 2 As for analysis of housing market using hedonic price function, refer to Linneman (1986), Ito and Hirono (1993), Hirono (2002, 2003), Nishimura and Shimizu (2002), Nakagami (1995). 3 The actual prices here contain transactor error. In other words, the actual prices are prices which are realized in the market if there is no disclosure of information.

4 Since I use the results in Table 3.2 for calculation of γ, I divide Pat, i  Pt, i by Patq, i which is the actual prices at time (t – q) to obtain Vi. 5 I cannot exclude the case of 1 < γ theoretically. However, I have preliminary result that 0 ≦ γ ≦ 1 in the housing market in Hirono (2002, 2003).

3.2 Housing Valuation Method Using the Hedonic Price Function

35



Pat, i  Pt, i : Pat, i  P t, i ¼ 1 : γ. In other words, 1  γ shows the proportion of error committed when estimating the hedonic function to the total error, for this model. Ri, tq—the q-period return on houses i between time t and time tq—is defined  a a a by Pt, i  Ptq, i =Ptq, i : Here, Ri, tq is the appreciation return. Ri, tq

¼ ¼



 Pat, i  Patq, i =Patq, i
a a a  a Pat, i  P t, i =Ptq, i  Ptq, i =Ptq, i þ Pt, i =Ptq, i

ð3:4Þ

a ¼ γ V i  1 þ P t, i =Ptq, i :

I calculate γ by regressing R i, tq on Vi in Eq. (3.4). Patq, i and P t, i can be treated as given in Eq. (3.4). Here, return is the appreciation return. I exclude the analyst error from the total error by regressing the appreciation return on Vi. In other words, I use the nature that the actual change in the appreciation return is γVi in Vi. Thus, I do not include rent or mortgage payments in the return on the houses. Linneman (1986) calculates the ratio of the transactor error in the total error by regressing the appreciation return on Vi. However, Linneman (1986) adopts the return on the houses from time t to time t+3 as the appreciation return to investigate the efficiency of the housing market ex-post (after 3 years). I use the return on the houses from time tcq to time t (i.e., up to now), since the purpose of this chapter is to develop a housing valuation method at the present period. I add the two equations given below to the model of Linneman (1986). Since the real theoretical price of house P t, i does not include the analyst error, the difference
between the actual Pat, i and real theoretical (P t, i ) housing prices is the transactor error. Thus, I have the following equation:
a TRi ¼ Pa t, i  P t, i =Ptq, i

ð3:5Þ

Here, I express TRi in proportion to Patq, i to make it consistent with the expression of Vi . Equation (3.6) does not change, even when TRi and Vi are not expressed in proportion to Patq, i Eqs. (3.3) and (3.5) yield the following equation: P t, i

¼ Pat, i  γV i  Patq, i


¼ Pat, i  γ Pat, i  Pt, i :

ð3:6Þ

I obtain the real theoretical housing using Eq. (3.6) and value the price of house i at time t at level P t, i .

3 Appropriate Housing Valuation Using Hedonic Price Function and. . .

36

3.3

Theoretical Value of Housing Price

This chapter extends Rosen’s (1974) model because it allows for describing the hedonic price function. Rosen’s (1974) model expresses the theory of the hedonic price function as a problem of determination of both consumption and supply. This model is extended through various equations presented in Sect. 3.2. This section mainly shows the nature of the equilibrium price of house P t, i : Rosen (1974) describes the variables of the utility and cost functions not as the volume but attributes of goods. According to Rosen (1974), prices expressed by the hedonic function are optimal in the sense that consumers maximize the profit and producers minimize their cost. In addition, the hedonic price Pt(z, ηt) is the equilibrium price of the house with the attributes denoted as z as follows. Location of a house with attribute z in the space of attributes is described by the vector of coordinates z ¼ (z1, Z ) where Z ¼ z2, . . ., zn. zj is the amount of the jth attribute of the house. I assume perfect competition and free entry of producers in the housing market. Consumers place positive marginal valuation on each attribute. Pt(z, ηt) is assumed to have continuous second derivatives. Pt in Eq. (3.2) (Sect. 3.2) that is the equation to be estimated and Pt(z, ηt) in this section are the same. However, the theoretical price of house calculated by the estimated Eq. (3.2) is different from the real theoretical price—the equilibrium price on Pt(z, ηt) in Fig. 3.3—by the analyst error. I express the parameter of consumer taste as αt. Let vt denote all the other goods consumed. The utility function is written as U(vt , z1, Z; αt). In line with Rosen (1974), I assume that the utility function has usual properties and is strictly concave. Consumer is expected to buy one unit of the house with attribute z. I normalize the price of vt to 1 and denote the income as mt. The consumer problem is formulated as the maximization of utility subject to the budget constraint mt ¼ vt + Pt(z, ηt). The first-order condition is ∂Pt/∂zj ¼ Pj ¼ Uz j /Uv , j ¼ 1, 2, . . ., n. I assume that Pt(z, ηt) is convex. I prove the pareto optimality of the equilibrium in this section, and the second-order conditions are satisfied as seen in that part. I define the value function θ (z1, Z; u, mt, αt) as the amount a consumer is willing to pay for (z1, Z ) at a given level of utility u, income, and taste. In other words, this yields the following expression: U ðmt ‐θ, z1 Z; αt Þ ¼ u:

ð3:7Þ

Differentiate Eq. (3.7) to obtain Eq. (3.8) θzj ¼ U zj =U v > 0, θu ¼ 1=U v < 0, θm ¼ 1,

ð3:8Þ

θ is the increasing function of zj at a decreasing rate which can be explained by Eq. (3.8), and θzjzj < 0. θzj is the demand price for one unit increment in zj. Next, let Mt(z) denote the production of a firm constructing the house with attribute z. Each producer is assumed to produce a house with one type of attribute.

3.3 Theoretical Value of Housing Price

37

The total cost function of a producer is C (Mt, z; βt) where βt is the shift parameter of the cost function. βt changes with a change in the factor prices and production technology. I assume that C is convex and C(0, z) ¼ 0, CM > 0, Czj > 0. There are no indivisibilities in production. The marginal cost is positive and an increasing function of Mt. It is assumed that the marginal cost of each attribute is positive and a non-decreasing function of each attribute. Producers maximize profit π t ¼ MtPt(z, ηt)C(Mt, z1, Z; βt) by choosing Mt and z. The first-order conditions of profit maximization are Eqs. (3.9) and (3.10). Pt j ðz, ηt Þ ¼ Cz j ðM t , z1 , Z; βt Þ=M t ,

j ¼ 1, 2, . . . , n

Pt ðz, ηt Þ ¼ CM ðM t , z1 , Z; βt Þ:

ð3:9Þ ð3:10Þ

At the optimal value, the marginal revenue of an additional attribute equals its marginal cost of production per unit. The revenue from one unit of house equals the marginal cost of that house. I assume that the second-order conditions are satisfied. The offer function φ( z1, Z; π t, βt) is defined as the unit price of a house a firm is willing to accept at a certain level of profit when quantities produced are optimal. I have π t ¼ M t φ  C ðM t , z1 , Z; βt Þ,

ð3:11Þ

CM ðM t , z1 , Z; βt Þ ¼ φ:

ð3:12Þ

Differentiation of Eqs. (3.11) and (3.12) gives φzj ¼ Czj/Mt > 0 and φπ ¼ 1/ Mt > 0. I assume φz jz j > 0: I show below consumer producer, and market equilibriums when the information is complete. According to Rosen (1974), Pt(z, ηt) is the minimum market price a consumer should pay in the market. Utility is maximized when θ (z; u, mt, αt) ¼ Pt(z, ηt), θzj(z; u, mt, αt) ¼ Pt j(z, ηt) where z and u are the optimal values. In other words, the consumer equilibrium is the point at which Pt(z, ηt) and θ( z; u, mt, αt) are tangent to each other (Fig. 3.1). In Fig. 3.1, Pt(z, ηt) and θ (z; u, mt, αt) are depicted on the θ  z1 surface, which is cut at Z (where Z¼ z2, . . ., zn). There are many consumers with different αt. In Fig. 3.1, the value function of a consumer with taste α1t is θ1, and the value function of another consumer with taste α2t is θ2. The equilibrium of all the consumers is shown as a family of value functions. The envelope of the value functions represents the hedonic price function in the market.
Concerning the production
decision, I have Pt ðz , ηt Þ ¼ φ z1 , Z  ; π t , βt and Ptj ðz , ηt Þ ¼ φz j z1 , Z  ; π t , βt , given that Pt(z, ηt) is the maximum market price obtainable for the house with attribute z. In Fig. 3.2, the producer
equilibrium is depicted by the tangency between Pt(z, ηt) and φ z1 , Z  ; π t , βt : In Fig. 3.2, the z1  φ surface cuts at the optimal values of the other attributes. This figure also shows that the offer function of a firm with the shift parameter β1t is φ1, and the offer

38

3 Appropriate Housing Valuation Using Hedonic Price Function and. . .

Fig. 3.1 Equilibrium of consumers

Fig. 3.2 Equilibrium of producers

function of another firm with β2t is φ2. There is a distribution of βt across all the firms. The producer equilibrium is shown as the family of offer functions. The envelope of the offer functions represents the hedonic price function in the market. Figure 3.3 shows the market equilibrium. In the equilibrium, the value function of a consumer and the offer function of a producer are tangent to each other, and the common gradient of the value function and the offer function at the equilibrium is the gradient of the hedonic price function Pt(z, ηt) that clears the market. Thus, Pt(z, ηt) is the joint envelope of the value and
offer functions. In Fig. 3.3, the equilibrium price of house i with attribute z1 , Z  is P price, the t, i ; at this market
  demand price equals the supply price of houses with attribute
z1 , Z . The equilibrium price of house j with an attribute different from z1 , Z  is not P t, i : However, it is the coordinate where the hedonic price function is tangent to
both the value and the offer functions, as in the case of house i with attribute z1 , Z  . According to Rosen (1974), the equilibrium price P t, i of houses with attributes
  z1 , Z is the price evaluated by the hedonic price function (as for houses with this

3.3 Theoretical Value of Housing Price

39

Fig. 3.3 Market equilibrium

attribute). The housing price fully reflecting the information and without analyst error in Sect. 3.2—the real theoretical price of house P t, i in Eq. (3.6)—is the price evaluated by the hedonic price function (without analyst error). Hence, P t, i in Eq. (3.6) is equivalent to this equilibrium price. The equilibrium price in this theoretical model is equivalent to the real theoretical price of a house because of the nature of P t, i :

Next, I prove pareto optimality of the equilibrium E z1 , Z  , P t, i in this model, wherein the utility function is differentiable and concave in the non-negative quadrant. vt + Pt(z, ηt) is differentiable and convex in the non-negative quadrant. Thus, the second-order conditions of the consumer problem are satisfied. (I apply the KuhnTucker sufficient theorem for the non-linear maximization problem). Since the budget constraint is an equality constraint, I have a non-satiation hypothesis in the consumer problem. Moreover, based on the assumption, the second-order conditions of the producer problem are satisfied. This is the
model
of perfect competition. From those, the equilibrium in this model E z1 , Z  , P t, i is pareto optimal. It must be noted that, even if the model is for the perfect competitive market, an allocation associated with equilibrium may not necessarily be pareto optimal. For example, if the non-satiation hypothesis in the consumer problem is not satisfied, marginal utility takes the value zero and the equilibrium is not pareto optimal. Given this, the fact that hedonic price is a perfectly competitive price may not necessarily lead to a pareto optimal allocation. One of the features of the hedonic price function model is that marginal evaluation of each attribute of a good is positive, and we compare the marginal evaluation of each attribute. Therefore, in this chapter, the hedonic price function model assumes the positive marginal evaluation of each attribute, which leads to a positive marginal house valuation. Thus, consumption is not satiated. The other conditions of pareto optimality are satisfied as above. Therefore, the allocation associated with the equilibrium is pareto optimal. At this pareto optimal equilibrium, I have Pt j ðz , ηt Þ ¼ θZ j ðz ; u , mt , αt Þ ¼
φz j z1 , Z  ; π t , βt : In other words, as shown above, the gradients of the three

40

3 Appropriate Housing Valuation Using Hedonic Price Function and. . .

Fig. 3.4 Case of selling a house at a price lower than the market price

functions—the hedonic price function Pt(z, ηt) the value function θ(z; u, mt, α
t), and

   the offer function φ z1 , Z ; π t , βt —are the same at the equilibrium E z1 , Z  , P t, i

in Fig. 3.3. Thus, at equilibrium E z1 , Z  , P t, i in Fig. 3.3, a producer (a consumer) cannot be made better off without making a consumer
(a producer) worse off. For example, if the price of the house with attribute z1 , Z  is higher than P t, i then the producer would be better off and the consumer would be worse off. Next, I show the cases of the transaction error. When there is imperfect information, the actual prices of houses differ from their equilibrium prices in Fig. 3.3. Figure 3.4 shows the case in which imperfect information drives the producer to sell a house at a price lower than the market price. In this case, since the housing price the producer is willing to accept at the given level of profit is lower than that when the producer has
complete information, the offer function of the producer is
φ0 z1 , Z  ; π t , βt not φ z1 , Z  ; π t , βt in Fig. 3.4. When the attributes of houses are the same, a consumer
prefers buying a cheaper house. Thus, the price of a house j with attribute z1 00 , Z  will be Pat, i : The transactor error (Eq. (3.5)) will occur when 00 there is a difference between Pat, i and the real theoretical price of house P t, i : It must be noted that, if sellers undervalue their houses, the profits accrued to buyers
with complete information will correspond to the difference between the actual Pat, i and
00 real theoretical P house prices. t, i The sellers of used houses who are not professional real estate agents often have imperfect information. These sellers also undervalue their houses when making a rushed sale owing to a job relocation or a decision to move into the home of an adult child or nursing home. These rushed sales might lead to sellers to trade at a lower price than the real theoretical prices. Moreover, worse financial condition of builders might lead sales of houses at cheaper prices than the real theoretical prices. This deviation from the hedonic price may occur when a housing valuation method used is the sales comparison, the profit return, or the cost methods instead of the hedonic approach. Similarly, a buyer with imperfect information pays more than that

3.5 Disclosure of Information

41

required in the market can be explained by an upward movement of the value function θ( z1, Z; u, mt, αt) that leads to a transaction error.

3.4

Appropriate Housing Valuation Method

This section shows how to calculate the real theoretical prices of houses—the appropriate level of valuation of houses—by regressing the hedonic price function, estimating γ, and Eq. (3.6).

3.4.1

Estimation of the Regression Equation, γ, and the Real Theoretical Prices of Houses in the Dataset

I regress Eq. (3.2) using the dataset compiled from the prices and attributes of houses. The real theoretical prices are calculated in the following way. First, I calculate the theoretical prices of houses (in natural log) by substituting the attributes z1, z2, . . ., zn and yearly dummies in the estimated equation of hedonic price function, that is, Eq. (3.2). I change these prices by the exponential function and let the outcome be the theoretical prices of houses Pt, i : I calculate the difference between Pt, i and the actual prices of houses Pat, i and divide the outcome by Patq, i , that is,
V i ¼ Pat, i  Pt, i =Patq, i : Moreover, I estimate γ by regression of the appreciation return on houses on Vi. Since I already have Pat, i and Pat, i  Pt, i : I can calculate the real theoretical prices of houses P t, i from Eq. (3.6). In other words, the dataset provides γ and the regression equation, which are used to calculate the real theoretical price of houses. I propose to value houses at the real theoretical price of houses
P t, i :

3.5

Disclosure of Information

Although Japan has over 62 million houses, it has 54 million households (Ministry of Internal Affairs and Communications, 2018). Japan has less circulation of used houses than that of the United States and European nations. It is also difficult to make these house compatible with the needs of people at different stages of aging. For example, it may be helpful for the elderly to move from a detached house to a barrierfree rental house with elder care potential. This scenario has led to the question of the supply of detached houses, which may be suitable for some younger family with children. Given this, the circulation of used houses has become a political issue in Japan.

42

3 Appropriate Housing Valuation Using Hedonic Price Function and. . .

In this regard, the use of disclosure of information is critical to realize the adequate valuation and hence the circulation of used houses. Concerning new houses, the appropriate valuation of new houses and information disclosure are crucial to develop a high-quality, newly built housing market so that consumers can buy them without worry.

3.5.1

Information to Disclose and Effects of Information Disclosure

To realize the valuation of houses outside the aforementioned dataset used for estimating the regression equation, I propose disclosure of the following information: γ (parameter used to calculate the real theoretical prices of housing using Eq. (3.6)), the estimated equation of hedonic price function (Eq. (3.2)), and Eq. (3.6) by the government. Information for each house (i.e., the offer prices and attributes) is disclosed by magazines of housing and real estate agents. I also recommend the use of the Internet for facilitating rapid information disclosure.  The use of disclosed information makes it possible to calculate P t, i and then Pt, i using Eq. (3.6). This can lead to the realization of optimal housing prices in the housing market, as shown in Sect. 3.3. As stated earlier, information disclosure through a housing policy can provide an objective valuation of houses, which accelerates the circulation of houses in the market of used houses and fosters the market of newly built houses. The information disclosure can produce several effects. Firstly, the housing prices will be equilibrium prices P t, i at which the utility of consumer is maximized and the cost of producer is minimized. Secondly, the housing prices will be pareto optimal. Thirdly, according to Chap. 2, the level of housing ownership differs by 12.6–27.6% of the theoretical prices owing to the difference in information available to buyers. Since the proposed disclosure of information can realize P t, i which is the price in the case of complete information, it can contribute toward correcting the disparity in the level of housing ownership.

3.5.2

Application of the Method to Realize the Real Theoretical Prices of Houses

I show an application of this valuation method. Between the home seller and buyer, the price set at a given level, which includes the transactor error. Next, the seller resets the price of this house to P t, k using information of γ, the estimated regression equation, and Eq. (3.6).

3.6 Empirical Estimates

3.6

43

Empirical Estimates

This section presents an application of the proposed method to an actual dataset. I compile data of houses listed for sale in the Shukan Jutaku Joho, which is a weekly house information magazine. I calculate the theoretical prices of the houses by applying the hedonic approach in Sect. 3.2. I change these values by using the exponential function to find P t, i and calculate the differences between the actual and the theoretical prices of houses. Subsequently, I estimate γ by regression of Eq. (3.4). I value housing prices using Eq. (3.6).

3.6.1

Estimation of the Theoretical Prices of Houses

The theoretical prices are estimated for the period 1986–1993. Concerning this period, Chap. 2 indicates that not a few housing market traders from Tokyo had imperfect information because of large variability of housing prices. The sample comprises houses listed in the first and second week of the calendar year (which are usually sold in the third and fourth week of January). I use housing units along the Yamate line, the Chuo line, the Toyoko line, the Sobu line, the Keihin Tohoku line, and the Joban line, in the Tokyo metropolitan area. I set q ¼ 3. Since I calculate appreciation return using prices of houses at time t and at the preceding 3 years, I only include same type of condominiums listed at time t and at the preceding 3 years in the Shukan Jutaku Joho. This approach is different from that in Chap. 2, where I only include houses listed at time t and at 3 years later. The same condominiums means that they must have the same attributes—name of the condominiums, address, floor space, balcony area, facing direction, floor plan, and construction. The sample includes 323 houses in high-rise, residential condominium units (excluding dormitories, corporate housing, offices, cheap apartments, and shops).6 I use this dataset to estimate the hedonic price function of Eq. (3.2) and present the results in Table 3.1. Concerning the attributes in Table 3.1, the commuting time to Tokyo, Otemachi, or Hibiya stations is expressed in minutes, which is the sum of minutes on foot or by bus and train. The other statistically significant attributes are floor space (in square meters), building age, and dummy variables which are 1 when a condominium is along the Sobu line, the Joban line, and the Keihin Tohoku line, respectively (Table 3.1, Spec (a)).7 I try two different specifications to check the robustness of the result. Specs (b) and (c) include dummy variables indicating whether the condominium is constructed using prestressed concrete (PC) or whether it possesses a small garden, 6

Detached houses are not used as samples, since there were no detached houses which were listed again in Shukan Jutaku Joho 3 years later. 7 Toyoko Line Dummy and Chuo Line Dummy are insignificant. As for these dummies about lines, dummies are based on Yamate Line (i.e., as for the Yamate Line, Dummy ¼ 0).

44

3 Appropriate Housing Valuation Using Hedonic Price Function and. . .

Table 3.1 Estimated hedonic price function of houses for sale Spec Constant Commuting time Floor space Building age Sobu line dummy Joban line dummy Keihin tohoku line dummy Small garden, roof garden PC R SEE 2

(a) 7.436 (94.478) 0.015 (14.545) 0.018 (15.575) 0.009 (2.696) 0.551 (14.173) 0.654 (17.153) 0.516 (13.382)

(b) 7.436 (94.534) 0.015 (14.537) 0.018 (15.423) 0.009 (2.573) 0.549 (14.105) 0.651 (17.011) 0.516 (13.376) 0.086 (1151)

0.830

0.830

0.235

0.235

(c) 7.430 (94.738) 0.015 (14.344) 0.018 (15.530) 0.008 (2.480) 0.544 (13.975) 0.649 (17.044) 0.519 (13.503) 0.140 (1.884) 0.831 0.234

Notes: t-statistics in brackets Small garden, roof garden: a dummy variable ¼ l if condominium is with a small garden, two-sided balcony, roof garden or sunroom PC a dummy variable ¼ 1 if construction of condominium is PC (prestressed concrete)

two-sided balcony, roof garden, or sunroom. I use yearly dummy variables for the period 1987–1993, in the regression equation in Table 3.1.8 The adjusted R squares and signs of coefficients agree with the theoretical prediction. These results show that the regressions are successful. According to the results in Spec (a) in Table 3.1, every additional 10-minute commuting time depreciates the value of condominiums by 15%. An increase in the floor space of 10 m2 leads to an 18% increase in the price. In this context, it must be noted that the value of a condominium depreciates 0.9% annually. If condominiums have a small garden, two-sided balcony, roof garden, or sunroom, they are valued 8.6% higher (Spec (b) in Table 3.1); PC condominiums are valued 14% less in Spec (c). I include all the attributes listed in the Shukan Jutaku Joho in the sample and tried them in regression Eq. (3.2). The attributes that were insignificant in regressions were the closet space, storage space availability, balcony area, availability of parking space, situated on the first floor (or not), use (non-use) of reinforced concrete (RC) for construction, condominium’s position (e.g., facing east or facing south), corner condominium (or not), recent renovation (or not), tiled (or not), along the local train stations (or not). I calculate the theoretical prices of houses (in natural log) by substituting z1 , z2 , . . . , zn and yearly dummies in the estimated equation of Spec (a) in Table 3.1. This is because all the coefficients of explanatory variables of Spec (a) are statistically different from zero at the 5% level. Let Vi denote the difference between the actual prices of houses Pat, i and the theoretical prices of houses (transformed by the exponential function) Pt, i divided by the actual prices Patq, i : 8

Yearly dummies are based on the year 1986 (i.e., as for data for 1986, a yearly dummy ¼ 0).

3.6 Empirical Estimates

45

Table 3.2 Regression of return on Vi

1

dRi, t  q/dVi 0.523*

t-statistics 8.867

R2 0.762

Note: *Statistically different from zero at the 5% level

3.6.2

Estimation of γ

I assume that q ¼ 3 and regress 3-year appreciation return on houses Ri, t3 on Vi in Eq. (3.4). Table 3.2 shows the result of the regression. Since there is a significant fluctuation in return on houses, I include yearly dummies as explanatory variables in the regression equation in Table 3.2. I also include a dummy showing that a condominium is recently renovated, given that renovation increases the appreciation return on houses. Table 3.2 shows that γ ¼ 0.523. In Table 3.2, the coefficient of Vi is significant. It must be noted that α (¼ – 0.527) in Chap. 2 and γ in this chapter are expected to have reverse signs, and the sign of γ is in agreement with this prediction.

3.6.3

Example of the Calculation of the Real Theoretical Prices of Houses

Firstly, I show the valuation method of houses in the dataset with which I estimated the regression equation of the hedonic price function. The actual price of a high-rise condominium in Omiya along the Keihin Tohoku Line was 26.5 million yen in 1993. The attributes of this condominium are 50-minute commuting time, floor space of 66 m2, and building age of 10 years and 2 months. By substituting these attributes and the yearly dummy to the estimated hedonic price function (a) in Table 3.1
and transforming by the exponential function, I find that the theoretical price Pt, i of this condominium is 31.80128 million yen. I obtain γ ¼ 0.523 by the regression of Eq. (3.4). Given this value and from Eq. (3.6), I can calculate the real theoretical price of this house: P t, i ¼ 29.27257 million yen. Using the same method, I assess the real theoretical price of a condominium in Ikebukuro along the Yamate line. The actual price of this condominium was 79.8 million yen in 1991. The attributes of the condominium are 24-minute commuting time, floor space of 50.66 m2, and
building age of 5 years and 3 months. I can calculate the theoretical price Pt, i of this condominium as 78.56205 million yen. Hence, I value this house as P t, i ¼ 79.15255 million yen. Next, I calculate the real theoretical price of a house outside this dataset. To this end, I use the following disclosed information: estimated equation (a) in Fig. 3.1, γ ¼ 0.523, and Eq. (3.6). The actual price of a condominium in Mitaka along the Chuo Line was 45.5 million yen in 1989. The attributes of the condominium are 39-minute commuting time, floor space of 54.45 m2, and building age of 8 years and 6 months. I substitute this information in (a) in Table 3.1. I change this outcome

46

3 Appropriate Housing Valuation Using Hedonic Price Function and. . .


using the exponential function to obtain the theoretical price Pt, i of this condo
minium which is 44.744 million yen. From (3.6), I have Pt, i ¼ 45.10461 million yen. I value this house at the level of P t, i . In another case, the actual price of a condominium in Funabashi along the Sobu line was 21.5 million yen. The attributes of the condominium are commuting time of 39 min, floor space of 55.68 m2, and building age of 9 years
and 4 months in 1988. I calculate the theoretical price of this condominium Pt, i as 26.97081 million yen. The value of this house is P t, i ¼ 24.36123 million yen. The need for disclosed information highlights the crucial role of the government. The government is assumed to try all the attributes mentioned in the Shukan Jutaku Joho, during the estimation period, in the regression Eq. (3.2). The government uses the dataset comprising these attributes and housing prices (at time t and tq). The government regresses Eq. (3.2) and discloses the regression equation containing only explanatory variables statistically different from zero at the 5% level. The government also estimates and discloses γ and Eq. (3.6). The government discloses this information through the Internet. This information disclosure can reduce the appraisal cost of housing incurred by real estate appraisers. Concerning the attributes and offer prices of each house, this information is disclosed to the housing market participants through magazines of house information or real estate agents.

3.7

Conclusion

This chapter proposed a housing valuation method using a hedonic price function without analyst error. It demonstrated that the housing valuation can be realized such that the consumers and producers act optimally. Pareto optimality in housing allocation can be achieved also by valuing houses using the hedonic price function and removing the analyst error. This proposed valuation method entails the estimation of the real theoretical prices of houses in a given dataset. In other words, we can obtain the real theoretical prices of houses in the dataset by regressing the hedonic price function and using the same dataset to estimate γ. This chapter also proposed the use of disclosed information on the Internet that accelerates the housing valuation method using hedonic price function. Disclosed information can allow for appropriately estimating the prices of houses outside the dataset. Concerning the future scope of this study, in line with Linneman (1986), this chapter assumed that the ratio of the transactor error to the total error is the same for all the houses. In the future study, I will relax this assumption. One practical notice is that the proposed method and disclosed information can be effective if used for housing valuation in a regional housing market such as the Tokyo metropolitan area. Hence, it may not be effective for the whole of Japan. This is because the valuation of the attributes of houses (coefficients of the hedonic price function) may differ among different regional housing markets.

References

47

References Griliches, Z. (1961) “Hedonic Price Indices for Automobiles: an Econometric Analysis of Quality Change,” in The Price Statistics of the Federal Government, General Series, No. 73, NBER, reprinted in Griliches (1971) Griliches, Z. (ed.) (1971) Price Indexes and Quality Change, Cambridge: Harvard University Press. Hirono, K. N. (2002) “Joho ga Jutaku Hoyu ni Oyobosu Eikyo,” Journal of Personal Finance and Economics, 17, 119–26. Hirono, K. N. (2003) “Joho no Fukanzensei to Jutaku Hoyu no Kakusa,” Jutakutochikeizai (Quarterly Journal of Housing and Land Economics), 49, 33–39. Hirono, K. N. (2004) “Appropriate Valuation of Housing Using a Hedonic Price Function and Disclosure of Information,” Studies in Regional Science, 34(1), 397–410. Hirono, K. N. (2012) Jutaku no Shitu ni Kansuru Jeizaibunseki, Tokyo: Tagashuppan. Ito, T. and K. N. Hirono (1993) “Efficiency of the Tokyo Housing Market,” Monetary and Economic Studies, 11(1), 1–32. Linneman, P. (1986) “An Empirical Test of the Efficiency of the Housing Market,” Journal of Urban Economics, 20(2), 140–54. Ministry of Internal Affairs and Communications (2018) Outline of the Housing and Land Survey, Tokyo. Nakagami, Y. (1995) “Fudosan Shijo niokeru Genzaikachimoderu nituite,” Jutakutochikeizai (Quarterly Journal of Housins and Land Economics), 16, 20–27. Nishimura K. and C. Shimizu (2002) “Shogyochifudosankakakushisu no Seido,” Quarterly Journal of Housing and Land Economics, 43, 28–35. Rosen, S. (1974) “Hedonic Prices and Implicit Markets: Product Differentiation in Pure Competition,” Journal of Polital Economy, 82(1), 34–55.

Chapter 4

Housing Loan Policy in Japan: Benefits and Drawbacks of Low-Interest Rate Housing Loans by Government Agency

Abstract This study quantitatively verifies whether the Government Housing Loan Corporation’s (GHLC) low-interest rate policy and the value of the GHLC’s housing loans prompted new housing development in Japan. To this end, it employs the vector autoregression model to analyze data on new housing development between 1981 and 2004. This study also identifies the drawbacks of the direct housing loan under GHLC. The GHLC housing loans suppressed the activities of private financial institutions and caused a budget deficit, thereby burdening the taxpayers. However, the loan policy was beneficial despite these drawbacks. The results show that the low-interest rate increased new housing construction to the extent where the GHLC’s budget exceeded the actual amount of the GHLC’s home loans. The findings are also that the amount of the GHLC housing loans was effective in increasing new housing construction to the point of structural change, after which the effect declined. These findings identify the policy’s contribution to the size of Japan’s housing stock. Keywords Housing loan · Low-interest rate policy · Vector autoregression model · Government agency

4.1

Introduction

The main objective of the Japanese housing loan policy was to increase the supply of housing and relieve the post-WWII housing shortage, which had arisen until a period of rapid growth, using direct housing loans provided by the Government Housing Loan Corporation (GHLC) (GHLC, 1993). In this chapter, I investigate the effect of the GHLC’s housing loan policy on housing construction in Japan and analyze the positive effects of the GHLC’s direct housing loans. On the other hand, there existed controversy resulting from the negative effects of these direct housing loans issued by the GHLC, which is the second topic discussed in this chapter. Firstly, I outline the transition of Japanese housing finance, focusing on the role of the GHLC. Secondly, I analyze the extent to which low-interest rate policy and the value of loans issued by the GHLC, which was responsible for the vast majority of © Springer Nature Singapore Pte Ltd. 2022, corrected publication 2022 K. N. Hirono, Economic Analysis of Housing Policy in Japan, New Frontiers in Regional Science: Asian Perspectives 64, https://doi.org/10.1007/978-981-19-4925-8_4

49

50

4

Housing Loan Policy in Japan: Benefits and Drawbacks of Low-Interest. . .

publicly issued housing loans, raised the supply of housing as the GHLC’s benefits. Thirdly, I investigate the drawbacks of direct housing loan of the GHLC, namely, oppression of commercial activity, the general account’s compensation for the deficit of the GHLC, and dead weight loss of the society caused by lending housing loan at interest rates lower than the market interest rate. Since the interest rates of GHLC housing loans and the value of loans issued by the GHLC are then used as policy variables, I verify quantitatively how the development of new housing is affected when the interest rate on GHLC housing loans is lowered that is whether or not the low-interest rate policy of the GHLC prompted the building of more homes and the effect of the value of GHLC’s housing loan on new housing development. Following this, I investigate whether the rate at which new houses were built by lowering the GHLC’s interest rate changed in the third quarter of 1999, when the GHLC’s budget exceeded the actual value of issued housing loans.1 Moreover, I find that a one-standard-deviation increase in the value of GHLC-issued housing loans increased new housing development by 8214.7 units up to the second quarter of 1999. Finally, I examine whether this effect has continued since the third quarter of 1999. I evaluated the GHLC’s housing loan policy by employing Vector Auto Regression (VAR) techniques, particularly the impulse response function. I chose this method for the following reasons: Firstly, by using an empirical model that analyzes house construction using OLS, endogenous variables can be included as explanatory variables and create bias. Secondly, the VAR model is one of the most suitable structures for identifying relationships in the real economy without imposing theoretical constraints. This is the first study that verifies quantitatively the effects of both the low-interest rate policy and the value of the GHLC’s housing loan by employing a VAR model. The study identifies the benefits and the drawbacks of the GHLC’s low-interest rate housing loans, including the effect on consumers and general welfare for the first time. Before beginning my analysis of the impulse response functions, I performed tests to check for seasonality, stationarity in the data, and the existence of cointegration between the variables. I establish that new housing development, the interest rate on GHLC housing loans, and the value of GHLC-issued housing loans are non-stationary series with unit roots. In addition, I clarify the nature of the data used in this research and use the variance decomposition technique to check the validity of the conclusions drawn from the analysis of the impulse response function. A number of authors have investigated the GHLC. Kamoike (1991) presents a theoretical model describing the behavior of households demanding housing and firms supplying housing, and analyzing GHLC loans. However, Kamoike (1991) investigates the effect of GHLC policy on housing prices and rents, rather than on housing construction, and his conclusion remains indeterminate.

1

Value of housing loans refers to total value of money lent for house purchases.

4.1 Introduction

51

Homma et al. (1988) investigate the effect of changes in the GHLC interest rate on housing demand. They estimate the demand function of houses owned by age group, and show that a lowering of the GHLC lending rate reduced the cost of capital and raised housing demand. The difference between Homma et al. (1988) work and this study is that Honma et al. do not investigate the effect of the GHLC’s interest rate policy on housing construction either directly or quantitatively. Even if housing demand does increase when interest rates are lowered, the effect of this on housing construction depends on the elasticity of the supply curve of housing stock. Moreover, Homma et al. (1988) use data for both 1977 and 1986. Since the purpose of this study is to evaluate the GHLC’s policy and examine changes in policy effects after the structural changes within the Japanese housing market had occurred, our periods of estimation differ. Yoshino and Nakata (2000) show that an increase in the value of public housing loans does stimulate housing investment, and that this effect has declined since the Heisei Recession of 1992. Their conclusion is as follows. Commercial activities can be suppressed by public housing loans; therefore it is desirable that only private financial institutions issue such loans. However, I can interpret their empirical findings by stating that GHLC housing loans still stimulated housing investment in 1988. Unlike Yoshino and Nakata (2000), I adopt the VAR method and the intention of analysis is not only the effect of the value of GHLC-issued loans for housing construction but also the impact of the GHLC interest rate. The estimation period also differs from Yoshino and Nakata (2000) because of this difference in focus. As demonstrated by the above, this study is the first to verify the effect of changes in the GHLC interest rate and of changes in the value of GHLC housing loans granted for housing construction. I show that the effects changed after the budget came to exceed the actual value of housing loans being granted by the GHLC. In addition, unlike the above studies, this analysis aims to reveal the underlying economic structure from the data. For the first time in the field of Japanese housing policy, Japanese housing development, and the Japanese housing market, time series analysis is employed—not only in the sense that time series data is used, but also in using time series models such as the VAR model and impulse response function. To date, few researchers have used VAR analysis to investigate housing policies. Using the impulse response function, Pozdena (1990) demonstrates that deregulation, such as in the removal of interest rate ceilings on deposit rates, weakens the linkage between the interest rate and housing construction in the United States. However, this research does not aim to identify the effect of policies implemented by a public financial institution. As for the failings of GHLC, Shiota et al. (1999) note the general account’s compensation for the deficit, and Izu (1999) and Moriizumi (1996) indicate the suppression of commercial activities. By contrast, this chapter notes the welfare loss which low-interest rate direct housing loans under GHLC created in the housing market. For the proposed privatization of two United States Government-sponsored housing enterprises—Fannie Mae and Freddie Mac—in 2019, the Japanese experience might offer some lessons: (1) The low-interest rate housing loans of the GHLC,

52

4

Housing Loan Policy in Japan: Benefits and Drawbacks of Low-Interest. . .

which is a government housing loan corporation, contributed to housing construction; (2) We cannot say that securitization of housing loans backed by the Japan Housing Finance Agency (JHFA, the successor of the GHLC), which imitated the securitization model of the above United States Government-sponsored housing enterprises, is successful; and (3) Housing loan business is not so profitable for private financial institutions during periods of low-interest rates. In Sect. 4.2, I provide an outline of the Japanese housing finance system. In Sect. 4.3, I specify the objectives and describe the method of analysis employed, identify the model and the data used, and check the properties of the data using unit root tests and cointegration tests, amongst others. I then investigate the impact of the GHLC’s policies on housing construction. Section 4.4 describes the drawbacks of direct housing loans under the GHLC. Section 4.5 concludes.

4.2

Overview of the Japanese Housing Loan System

In this section I provide an overview of the Japanese housing loan system and note its major transitions. I summarize actual situation of public housing loans and private housing loans in Japan. In addition, I provide a review of the abolition of GHLC as carried out by the Koizumi administration.

4.2.1

The Japanese Housing Loan System

The Japanese housing loan system consists of a public housing loan system that is managed by public institutions and a private housing loan system operated by multiple private financial institutions. The public institutions which issue housing loans are the GHLC and local public agencies, and the private institutions are city banks, local banks, and Shinkin banks. The total value of outstanding housing loans was ¥ 1,878,163 trillion at the end of the 2016 fiscal year, and the total value of new housing loans was ¥ 245,651 trillion (Housing Loan Progress Association, 2018). Table 4.1 illustrates the transition in the composition ratio of new housing loans as for public and private financial institutions. According to Table 4.1, public housing loans accounted for 50% of the total in 1993. This contrasts starkly with ratios around 20% in the United States, the United Kingdom, and Germany in 1996, demonstrating that the public housing loan system is much larger in Japan. Among public housing loans, the loan value for new housing under the GHLC reached 40% of the total in 1993. At that time, the Japanese Government decided to abolish both direct financing and the GHLC, thereby reducing direct financing of housing loans of the JHFA (which is GHLC’s successive organization) to zero by

4.2 Overview of the Japanese Housing Loan System

53

Table 4.1 Transition in the ratio of new housing loan amount (%) Fiscal year Public institutions GHLC Direct finance Loan through securitization Local public agency The rest Private institutions Domestic banks (City banks
Local banks
Trust banks
Trust accounts) Shinkin banks Labor banks Others Total

1993 49.9

1998 36.1

2003 11.0

2008 5.3

2013 9.6

2015 12.2

40.1 0.0 1.9 7.9 50.1 33.1

30.6 0.0 1.5 4.0 63.9 48.8

8.7 0.0 1.4 0.9 89.0 69.4

0.0 3.5 1.5 0.3 94.7 75.1

0.3 9.0 0.3 0.1 90.4 71.7

0.2 11.6 0.4 0.0 87.8 69.4

6.4 2.9 7.7 100.0

7.4 4.4 3.3 100.0

10.2 7.0 2.4 100.0

8.6 8.4 2.6 100.0

8.7 7.7 2.3 100.0

10.0 6.6 1.8 100.0

Note: These are figures of housing loans of individuals Source: Calculated by the author using data from JHFA (2017) and GHLC (2003)

2008. Since this time, the receivables from housing loans financed by the JHFA’s securitization project have been gradually increasing (see Table 4.1).

4.2.2

Public Housing Loan

The GHLC was responsible for the vast majority of publicly issued housing loans in Japan (Table 4.1). Therefore, this subsection provides a history of the GHLC and the succeeding JHFA, as well as specifying their modes of operation. The GHLC was founded in 1950, according to the plan of the General Head Quarter (GHQ), to issue housing loans and resolve the issue of housing shortages in Japan after WWII (GHLC, 1993). As a result, the GHLC played a major role in expanding the housing supply and improving the quality of housing in Japan during the postwar reconstruction period and the first half of the rapid growth period, while private financial institutions were being urged to supply industrial funding. The GHLC together with the public housing system and the Japan Housing Corporation (which became the Urban Infrastructure Development Corporation in October 1999) bore responsibility for housing policy in the postwar period. The GHLC was a part of the Fiscal Investment and Loan Program (Zaisei Tōyūshi) (FILP). The FILP was the system by which the Ministry of Finance collected money from Japanese people in the form of postal savings and pension savings. This money was used to finance special corporations, of which the GHLC was one. The operating interest rate on GHLC-issued loans was lower than the procurement interest rate of the FILP. The difference was covered by the general budget (ippan kaikei) of the Japanese Government.

54

4

Housing Loan Policy in Japan: Benefits and Drawbacks of Low-Interest. . . (2) mortgage loans

(4) MBS

(1) housing loans (6) payment (7) repayment

(8) repayment

(5) payment (9) principal and interest

entrust mortgage loans (4) collateral

Fig. 4.1 Scheme of operation for securitization of housing loans. Source: JHFA (2017)

The GHLC contributed to the enlargement of the housing stock in Japan. However, it was abolished as part of the structural reform implemented by the Koizumi administration in 2007 following criticism over its commercial activities and the coverage of its deficit by the general budget. The JHFA took over the institution and began operations of securitizing housing loans. Securitization was undertaken as follows (see Fig. 4.1): (1) Private financial institutions make housing loans to customers (debtors) at longterm fixed interest rates. (2) Private financial institutions sell mortgage loans to the JHFA. (3) The JHFA entrusts these mortgage loans to trust banks for security. (4) The JHFA issues mortgage-backed securities (MBSs) to investors, using mortgage loans as collateral. (5) The JHFA receives payment for MBSs from investors. (6) The JHFA pays private financial institutions for mortgage loans with this payment from investors. (7) Private financial institutions receive repayment (principal and interest) from customers. (8) Private financial institutions pass these repayments on to the JHFA. (9) The JHFA pays the principal and interest to investors, as for MBS. Table 4.1 shows that the value of purchase receivables obtained through the securitization of mortgage loans (lent by private financial institutions) accounted for 11.6% of the total value of new housing loans in 2015. Housing loans purchased in this way are managed by 327 private institutions under the name of “Flat 35” (Yasui, 2015). However, 71.4% of private financial institutions were found not to have experience of securitization and did not consider this necessary. Furthermore, 72.4% of private financial institutions attributed this phenomenon to a lack of knowledge regarding securitization (JHFA, 2017); in other words, the securitization of JHFAissued loans is not going smoothly. According to the JHFA (2017), about 80% of private financial institutions are willing to manage housing loans. These private financial institutions point to

4.3 The Benefits of Direct Housing Loans Issued by the GHLC

55

increases in the value of outstanding loans (73.4%), the enforcement of household transactions (71.3%), and the recent decrease of ratio of the loans to firms (33.2%) as reasons for this attitude. This shows that these institutions want to take on housing loans which allow them to earn long-term profits. On the other hand, private financial institutions note important points as matters of concern regarding the risks of housing loans. These factors include decreases in margins due to interest rate competition (95.5%), a medium- to long-term deterioration in profitability (60.6%), and housing loan refinancing to other private financial institutions (58.1%) (JHFA, 2017). Yasui (2015) demonstrates that housing loan businesses cannot be profitable in this low-interest environment.

4.3

The Benefits of Direct Housing Loans Issued by the GHLC

In this section, I explore the benefits of direct housing loans of the GHLC. I check the effects of low-interest housing loan of the GHLC on new housing development in Japan. Then I verify whether the value of housing loan of GHLC accelerated Japanese housing construction or not.

4.3.1

Method of Analysis and Preliminary Tests

In this subsection, I first outline the method of empirical analysis employed and specify the time series models estimated. I then describe the data used. As preliminary tests to check the nature of the data used in the VAR analysis, I examine seasonality and stationarity of the data and I also perform the cointegration tests. I test the effect of two policy variables on housing loan markets. I first analyze the effect of the GHLC’s interest rate on housing construction. I then analyze the effect of changes in the value of GHLC-issued housing loans on new housing development (see footnote 1). I identify periods when the budget exceeded the actual value of GHLC housing loans and test for structural changes thereafter, assuming that GHLC-issued housing loans become less effective as a policy instrument since then.2 During this time, the interest rates of private financial institutions were declining and many GHLC-issued housing loans were being refinanced by private financial institutions as the interest rates on short-term housing loans from these institutions dropped below the GHLC rate. I analyze the housing loans of individuals because the objective of this investigation is to determine whether this housing loan policy satisfied the housing needs of 2

The information was obtained from a contact in the housing loan market.

56

4

Housing Loan Policy in Japan: Benefits and Drawbacks of Low-Interest. . .

individuals, and because the majority of GHLC-issued loans were offered to individuals. I measure housing by unit rather than area in order to consider the housing stock as influenced by housing policy and because the Japanese Government and the GHLC use the number of units (houses) to measure housing.

4.3.1.1

The Model

The models developed use VAR systems. In an autoregression (AR) model, the current values of a variable are estimated using its past values. The VAR model is an extension of an AR model with several variables (Yamasawa, 2004). In general, the equations to be estimated in the VAR model are as follows: X t ¼ αx þ

n X

βxi X ti þ

i¼1

Y t ¼ αy þ

n X i¼1

n X

γ xi Y ti þ ux t

ð4:1Þ

γ yi Y ti þ uy t

ð4:2Þ

i¼1

βyi X ti þ

n X i¼1

That is, in the estimation of VAR equation systems, the current values of Xt and Yt are regressed onto lagged values of Xt and Yt. According to Yamasawa (2004), much of the research using quarterly data has employed a lag of a year (i.e., n ¼ 4). Thus, in this study, I also let n equal 4. I employ a two-variable VAR system, rather than a large VAR system with many variables, in order to only take into account the necessary number of variables for acquiring sufficient degrees of freedom (Yamasawa, 2004), but also because the scarcity of data relating to post-structural change in the Japanese housing market restricts the size of the VARs. Thus, my VAR system is as follows: X t ¼ αx þ βx1 X t1 þ βx2 X t2 þ βx3 X t3 þ βx4 X t4 þγ x1 Y t1 þ γ x2 Y t2 þ γ x3 Y t3 þ γ x4 Y t4 þ ux t

ð4:3Þ

Y t ¼ αy þ βy1 X t1 þ βy2 X t2 þ βy3 X t3 þ βy4 X t4 þγ y1 Y t1 þ γ y2 Y t2 þ γ y3 Y t3 þ γ y4 Y t4 þ uy t

ð4:4Þ

In Eqs. (4.3) and (4.4), X1 and Y1 change according to changes in X0, X2 and Y2 change according to the change in X0, X1 and Y1. In this way, a VAR system expresses mutual dependence between the variables. In this chapter, I estimate two VARs. In Eq. (4.1), where I examine the effect of a low GHLC interest rate policy on housing development, Xt in (4.3) and Yt in (4.4) refer to housing development and the interest rates of GHLC-issued housing loans.

4.3 The Benefits of Direct Housing Loans Issued by the GHLC

57

In Eq. (4.2), where I examine the effect of the value of GHLC-issued housing loans, these refer to new housing development and the value of GHLC-issued housing loans. The choice of VAR models can be explained as follows. By assuming a theoretical model and estimating this model by OLS, with housing construction included as a dependent variable, endogenous variables, such as the interest rate of housing loans, would often be included as explanatory variables. This would undermine the assumptions of OLS regression: that explanatory variables should either be exogenous variables or lagged endogenous variables. Equally, applying OLS to simultaneous equation systems, including housing development as an explanatory variable, produces bias. However, VAR models eliminate these issues, since no distinction is made between endogenous and exogenous variables. Moreover, VAR models are appropriate in this context, because this study investigates the effect of a shock from one variable to another using impulse response functions. It evaluates the GHLC policy by studying observed effects, rather than by constructing a theoretical model. Thus, VAR models, which capture the relationships between variables without theoretical constraints on equations, are more suitable.

4.3.1.2

The Method of Empirical Analysis

Before applying my VAR model, data processing must be undertaken. Firstly, I test the data for seasonality and compensate for seasonal trends in order to capture movements in the housing market without seasonal movements. Secondly, I make non-stationary series (with unit roots) stationary before estimating equations and analyzing through VAR models. To do this, I check for stationarity and difference non-stationary values to make them stationary. Thirdly, I examine the cointegration of series. Since for those that are cointegrated, I have to apply the vector error correction model (VECM) rather than a VAR model. In this chapter, the results of impulse response functions are analyzed. I study the effect of both interest rates and the value of GHLC-issued housing loans on new housing development. I use impulse response functions because they show the movements of other variables when a shock (impulse) is experienced by one variable. The impulse response functions are as follows. Assume that X1 ¼ X2 ¼ Y1 ¼ Y2 ¼ 0 in Eqs. (4.3) and (4.4) and suppose for simplicity that αx ¼ αy ¼ 0.3 Considering an impulse of ux0 ¼ 1 at time 0, such that X0 ¼ 1 and Y0 ¼ 0 at time 0. Substituting these into (4.3) and (4.4) produces X1 ¼ βx1 and Y1 ¼ βy1. Moreover,

3

No important change in the main argument of the study arises from this simplicity.

58

4

Housing Loan Policy in Japan: Benefits and Drawbacks of Low-Interest. . .

X 2 ¼ βx1 X 1 þ βx2 X 0 þ γ x1 Y 1 ¼ βx1 2 þ βx2 þ γ x1 βy1 , Y 2 ¼ βy1 X 1 þ βy2 X 0 þ γ y1 Y 1 ¼ βx1 βy1 þ γ y1 βy1 :

ð4:5Þ

The values can be calculated consecutively. X0, X1, X2, and X3 and so on are the response functions of X to an impulse to X; Y0, Y1, Y2, Y3 are response functions of Y to an impulse in X. Usually, the size of an impulse is one standard deviation of an error term. Unlike the simple comparison of coefficients, impulse response functions have the advantage of incorporating all direct and feedback effects into the model (Pozdena, 1990). For example, considering the change in housing development caused by an impulse to the interest rate of housing loans, the impulse response function first shows a change in housing development and then a change in the interest rate for housing loans caused by this change in housing development, which continues in a feedback loop. In addition, impulse response functions reveal the effect of a change in a variable more explicitly and visually. I also present the impulse response of housing development to a 1% change in the interest rate to show the effect in a more comprehensible way. I also use variance decomposition to identify the response of one variable to another and identify the proportion of the variation in housing development that can be explained by interest rate changes and that which can be explained by the value of GHLC-issued housing loans. I use this method because changes in these proportions suggest a structural change in the housing market (Pozdena, 1990).

4.3.1.3

Data

The study uses quarterly data from the period between 1981 and 2004,4 as this is the most time-disaggregated data available. I use the standard interest rate (kijunkinri) of the GHLC as the interest rate on GHLC housing loans (RKOUKO), and the value of new GHLC-issued individual housing loans as the value of GHLC housing loans (NEWLOAN). In this study, data on new housing development rather than data on housing investment is used, since the value of housing stock can be misrepresented due to problem in housing valuation when we calculate investment (Pozdena, 1990). For information on the source of the data, see the Appendix.

4.3.1.4

Seasonality

Next, I examine seasonality through the identification of periodic peaks and troughs in the time series data. However, where there are large fluctuations in values, seasonal peaks and troughs are difficult to distinguish from other fluctuations 4 To make the results of this chapter comparable with those of Hirono (1998), the starting point of my data is the same as that in his work.

4.3 The Benefits of Direct Housing Loans Issued by the GHLC

59

(Pyndick, 1981). Thus, I plot the original data to check for seasonality as a preliminary approach (the results are shown in Figs. 4.7, 4.8, and 4.9 of Appendix 1). By visually examining these figures, seasonality in the value of GHLC-issued housing loans (NEWLOAN) and new housing development (KOCHAK) can be identified, but not for the GHLC interest rate on housing loans (RKOUKO). Next, I calculate the sample autocorrelation coefficient of each series to check for seasonality. According to Table 4.2, new housing development (KOCHAK) exhibits annual seasonality. However, no definite patterns are observable in the other series. Moreover, these three series are found to be non-stationary in Sect. 4.3.1.5. Therefore, I examine the sample autocorrelation coefficient of the first-differenced series.5 As Table 4.3 shows, for the value of housing loans granted by the GHLC (NEWLOAN) and new housing development (KOCHAK), there are annual (i.e., four-quarter lag) peaks in the sample autocorrelation coefficient of the firstdifferenced series, while there are no evident annual peaks for the interest rate on GHLC housing loans (RKOUKO). Therefore, there is no significant seasonality in the interest rate on GHLC housing loans (RKOUKO) series. To reach a definite conclusion, I use F-tests in the process of X-12-ARIMA to check for seasonality. The results show that NEWLOAN and KOCHAK exhibit seasonality, while RKOUKO does not. From these tests, it is clear that the value of GHLC-issued housing loans and new housing development do exhibit seasonality, while the interest rate on GHLC housing loans does not.

4.3.1.5

Stationarity

To check for stationarity, I perform unit root tests known as the Dickey–Fuller (DF) tests. Table 4.4 shows the results of these tests. For each series, I produce three specifications: Specification 1, where the test regression does not include either a drift or a trend term; Specification 2, where the test regression includes only a drift term; and Specification 3, where the test regression includes both a trend and a drift term. I adopt the results in which the drift term and the trend term in the test regression of the random walk are significant at the 5% level. That is, I employ Specification 3 if both drift and trend terms are significant at the 5% level, Specification 2 if the drift term is significant but the trend term is not, and Specification 1 when both terms are not significantly different from zero at the 5% level. According to Table 4.4, the null hypotheses for the value of GHLC-issued housing loans (NEWLOAN), new housing development (KOCHAK), and the interest rate of GHLC housing loans (RKOUKO), that the series have unit roots at the 5% level, cannot be rejected. These series are therefore non-stationary at level.

5

As shown in Sect. 4.3.1.5, all the series are non-stationary. In such cases I looked at a sample autocorrelation coefficient of the first-differenced series to check seasonality (Wago, 1987).

60

4

Housing Loan Policy in Japan: Benefits and Drawbacks of Low-Interest. . .

Table 4.2 Sample autocorrelation coefficient of the original data

Lag k 1 2 3 4 5 6 7 8 9 10 11 12

RKOUKO 0.960 0.914 0.872 0.835 0.800 0.765 0.726 0.690 0.652 0.614 0.583 0.552

NEWLOAN 0.888 0.873 0.783 0.788 0.658 0.628 0.549 0.559 0.449 0.425 0.346 0.361

KOCHAK 0.565 0.391 0.365 0.587 0.222 0.067 0.048 0.280 0.010 0.137 0.112 0.159

Table 4.3 Sample autocorrelation coefficient of the firstdifferenced series

Lag k 1 2 3 4 5 6 7 8 9 10 11 12

RKOUKO 0.354 0.072 0.136 0.090 0.060 0.033 0.127 0.061 0.006 0.127 0.028 0.123

NEWLOAN 0.500 0.387 0.486 0.705 0.518 0.276 0.451 0.638 0.450 0.285 0.434 0.603

KOCHAK 0.297 0.174 0.288 0.677 0.241 0.157 0.290 0.600 0.188 0.174 0.282 0.638

Since NEWLOAN, KOCHAK, and RKOUKO are non-stationary at level, and since NEWLOAN and KOCHAK exhibit annual seasonality, I analyze the fourth (annual) difference of the original data. I then perform unit root tests (DF tests) on this differenced series. Table 4.5 indicates that the fourth-differenced series of NEWLOAN, KOCHAK, and RKOUKO are all stationary.

4.3.1.6

Cointegration Tests

Next, I run Johansen cointegration tests to identify whether there is cointegration between variables. For the VAR models, I make the series stationary by differencing if the series has a unit root. In addition, I include an error correction term in the model if variables display cointegration according to Yamasawa and Nakano (1998). The results of the cointegration tests show that there is no cointegration between new housing development and the interest rate of GHLC housing loans, or between new housing development and the value of GHLC-issued housing loans (Tables 4.6

4.3 The Benefits of Direct Housing Loans Issued by the GHLC Table 4.4 Unit root tests of the original series (in the level)

Series RKOUKO NEWLOAN KOCHAK

61

Test statistic 1.654 0.802 2.342

Specification 1 1 2

*

Statistically different from zero at the 5% level

Table 4.5 Unit root tests of the fourth-differenced series

Series RKOUKO NEWLOAN KOCHAK

Test statistic 3.194* 2.677* 2.778*

Specification 1 1 1

*

Statistically different from zero at the 5% level

Table 4.6 Trace test Series RKOUKO, KOCHAK NEWLOAN, KOCHAK

No. of cointegrations r (the null hypothesis) r¼0

Alternative hypothesis r≧1

Trace statistic 10.70

5% critical value 25.87

r¼0

r≧1

13.44

25.87

*

denotes rejection of the null hypothesis at the 5% level

Table 4.7 Maximum eigenvalue test Series RKOUKO, KOCHAK NEWLOAN, KOCHAK

No. of cointegrations r (the null hypothesis) r¼0

Alternative hypothesis r¼1

Maximum eigenvalue statistic 8.08

5% critical value 19.39

r¼0

r¼1

8.09

19.39

*

denotes rejection of the null hypothesis at the 5% level

and 4.7). These results are confirmed not only by the trace tests but also by the maximum eigenvalue tests. The inclusion of an error correction term is therefore not required in any VAR model.

4.3.2

Policy Evaluations

In order to evaluate the housing policy of the GHLC, I present the impulse response functions that show the reaction of housing development (KOCHAK) to changes in

62

4

Housing Loan Policy in Japan: Benefits and Drawbacks of Low-Interest. . .

the interest rate on GHLC housing loans (RKOUKO) and the value of new GHLCissued housing loans (NEWLOAN), which are both policy variables of the GHLC.6 According to the activities, plans, and performance list of GHLC financing agreements (GHLC, 2003), the budget exceeded the actual value of individual housing loans issued by the GHLC in the fiscal year of 2000.7 However, since the budget is indicated at the end of the fiscal year, I use the interpolation method to adjust according to quarterly data. The interpolation shows that the budget was surpassed in the third quarter of 1999. Therefore, I separate impulse responses up to the second quarter of 1999, when the budget for GHLC-issued individual housing loans was below the actual value, and since the third quarter of 1999, when the budget surpassed the actual value. The fact that the value of GHLC-issued housing loans was underbudget meant that the GHLC could no longer deliver effective housing loan policies.8 In order to make the KOCHAK, NEWLOAN, and RKOUKO variables stationary, I take fourth difference of the data, removing the annual seasonality of KOCHAK and NEWLOAN. I do not seasonally adjust my data using X-12-ARIMA because this method removes the seasonality of the original data by converting to moving averages several times. In this case, even if there is an apparent causal relationship in the original data, there are many occasions in which the causal relationship disappears or a spurious causal relationship could be identified instead (Wago, 1987). In other words, X-12-ARIMA can change the properties of the data.

4.3.2.1

The Impulse Response Function

(1) The effect of the GHLC interest rate on housing development I first use the impulse response function to investigate the effect of the interest rate on GHLC-issued housing loans on new housing development. Figure 4.2 shows the response of new house building to a shock in the interest rate (RKOUKO) up to the second quarter of 1999, when the budget for the GHLC individual housing loans was less than the actual value of debt issued. Figure 4.3 shows the effect of an interest rate shock on housing development since the third quarter of 1999, when the budget for the GHLC individual housing loans exceeded the actual value of loans issued.

6

Although policy variables of the GHLC include the term of limitation of housing loans, there is no possibility that the GHLC used this as a policy variable to increase or decrease housing starts. Therefore, I did not include the term of limitation of housing loans in my analysis. 7 GHLC individual housing loans were the total loan debt issued by the GHLC for building new owner-occupied houses, buying good-quality houses for sale, or for buying houses, condominiums, ready-built houses, reused houses, residential portions of urban inhabitability recovery plans, and urban redevelopment dwellings from the Corporation, according to my interview with GHLC officials. 8 As for the period up to the second quarter of 1999, demand for housing loans surpassed the budget of the GHLC, and the GHLC supplied loans in excess of the budget.

4.3 The Benefits of Direct Housing Loans Issued by the GHLC

63

New Housing Development (Units)

16000 12000 8000 4000 0 -4000 -8000

1

2

3

4

5

6

7

8

Quarters

Fig. 4.2 Impulse response of new housing development to a shock in the interest rate of the housing loans of the GHLC (–the second quarter of 1999)

New Housing Development (Units)

12000

8000

4000

0

-4000

1

2

3

4

5

6

7

8 Quarters

Fig. 4.3 Impulse response of new housing development to a shock in the interest rate of the housing loans of the GHLC (the third quarter of 1999–)

64

4

Housing Loan Policy in Japan: Benefits and Drawbacks of Low-Interest. . .

Figures 4.2 and 4.3 indicate the extent to which new housing development changed by a one-standard-deviation (of error term) shock in the GHLC interest rate.9 The horizontal axis of Fig. 4.2 measures the number of quarters following the initial shock. Figure 4.2 shows the impact on new housing development following the positive shock given in the first quarter. It displays a decrease of approximately 7000 units over two quarters.10 Figure 4.2 shows that, up to the second quarter of 1999, a positive shock to the interest rate on GHLC housing loans lowered housing development after two quarters. That is, the GHLC would increase new housing development by lowering the interest rate of its housing loans (RKOUKO). However, since the third quarter of 1999, as shown by Fig. 4.3, the impulse response adopts a zigzag shape, indicating an inconsistent effect of the GHLC interest rate on housing loans (RKOUKO) on new housing development over time.11 Table 4.8 presents the results specified in Figs. 4.2 and 4.3. In addition, it includes the impulse response of new housing development to a 1% change in the interest rate of GHLC housing loans (RKOUKO), where a 1% change signifies a change in interest rate from 1% to 2%, and not a change in interest rate from 1% to 1.01%. I also include the accumulated impulse response to examine the effect of the policy. According to Table 4.8, a positive shock of one standard deviation of the error term of RKOUKO lowered new housing development by 6770.7 units in six quarters (five quarters after the shock) up to the second quarter of 1999. In other words, a onestandard-deviation decrease in RKOUKO resulted in an increase in new housing development by 6770.7 units. In short, the low-interest rate policy of the GHLC had a positive effect on new housing development in this period. For the period since the third quarter of 1999, the effect of the same positive shock to RKOUKO on housing development over the six quarters was a decrease of 2208.9 units. Thus, the absolute value of response was lower than before the second quarter of 1999. In addition, between one to eight quarters after the shock, the impulse response took both positive and negative values. A 1% decrease in RKOUKO caused an increase in housing development of 26,291.7 units over six quarters up to the second quarter of 1999, whereas since the third quarter of 1999 the absolute value of an impulse response to a 1% decrease in RKOUKO decreased largely to 9237.4 units. This impulse response took both positive and negative values during the eight quarters following the shock. The accumulated impulse response up to the second quarter of 1999, as shown in Table 4.8, shows that a one-standard-deviation decrease in the interest rate of GHLC 9

In the studies of policy evaluation, variables in the model are often logarithmically transformed and the percentage change in a variable of policy object caused by a 1% change in policy variable is investigated. However, “the percentage change of interest rate which is already expressed in a percentage” is misleading. Thus, I did not logarithmically transform my series as did Pozdena (1990), who analyzed the effect of the TB rate on housing starts. 10 The method of analysis for Figs. 4.3, 4.4, and 4.5 is the same as Fig. 4.2. 11 Data and analysis in Sects. 4.2 and 4.3 of this chapter are from Hirono (2005, 2020). I have added analysis of the drawbacks of direct finance under the GHLC in Sect. 4.4.

Quarters since shock 1 2 3 4 5 6 7 8

Impulse response (1std dev.) –1999Q2 1999Q3– 0.0 0.0 1306.1 3272.2 6874.7 1597.7 7536.2 2094.4 7033.9 2035.7 6770.7 2208.9 4285.8 2387.5 2193.9 1191.2

Impulse response (1%) –1999Q2 1999Q3– 0.0 0.0 5071.6 13,683.7 26,695.7 6681.3 29,264.2 8758.5 27,314.0 8512.8 26,291.7 9237.4 16,642.5 9984.0 8519.1 4981.5

Accumulated impulse response (1std dev.) –1999Q2 1999Q3– 0.0 0.0 1306.1 3272.2 5568.7 1674.5 13,104.9 3768.9 20,138.8 5804.6 26,909.5 3595.7 31,195.3 5983.2 33,389.2 4791.9

Accumulated impulse response (1%) –1999Q2 1999Q3– 0.0 0.0 5071.6 13,683.7 21,624.1 7002.3 50,888.3 15,760.8 78,202.2 24,273.7 104,493.9 15,036.3 121,136.4 25,020.3 129,655.5 20,038.9

Table 4.8 Impulse response of new housing development caused by a shock to the interest rate of the GHLC housing loans

4.3 The Benefits of Direct Housing Loans Issued by the GHLC 65

66

4

Housing Loan Policy in Japan: Benefits and Drawbacks of Low-Interest. . .

housing loans increased new housing development by 33,389.2 units over eight quarters. Since the quarterly average of new house constructions was 198,546 units in Japan, this increment in new housing development amounts to 16.8% of the total. A 1% decrease in the interest rate of housing loans caused a rise in new house building of 129,655.5 units over the eight quarters. As for the period following the third quarter of 1999, a one-standard-deviation decrease in RKOUKO lowered new housing development by 4791.9 units, and a 1% decrease in RKOUKO reduced new housing development by 20,038.9 units. From these results, it can be verified that up to the second quarter of 1999, when the budget was lower than the actual value of individual housing loans issued by the GHLC, a policy of lowering the interest rate of GHLC housing loans had a positive effect on housing construction. Since the third quarter of 1999, when the budget exceeded the actual value of GHLC-issued individual housing loans, the effectiveness of any lowering of the interest rate vanished. These results show that the GHLC succeeded in increasing housing construction through its low-interest-rate policy up to the second quarter of 1999. During this period, the GHLC granted housing loans at a low rate of interest, and there was excess demand for the GHLC housing loans (of qq0 , as illustrated in Fig. 4.10 of Appendix 2). Faced with this excess demand, the GHLC increased the actual value of housing loans granted. Lowering the interest rate on GHLC housing loans increased the level of demand. Thus, the actual value of housing loans increased and housing construction rose as a result. Additionally, over the period when the budget exceeded the actual value of housing loans issued by the GHLC (i.e., since the third quarter of 1999), the effect of the GHLC’s low-interest-rate policy disappeared. The budget surpassed the actual value of loans because of competition with private financial institutions that were offering three-year housing loans at a lower rate of interest than the GHLC. As a result, an increasing number of consumers preferred the floating-rate housing loans of private financial institutions to the longer-term fixed-rate housing loans of the GHLC. Thus, the lowering of interest rates by the GHLC became largely ineffective. (2) The effect of the value of GHLC-issued housing loans on new housing development Figures 4.4 and 4.5 and Table 4.9 demonstrate the impulse response function on housing development from a change in the value of GHLC-issued housing loans. Figure 4.4 shows the effect of a one-standard-deviation shock in the residuals for the value of GHLC-issued housing loans (NEWLOAN) on housing development up to the second quarter of 1999, when the budget was less than the actual value of GHLCissued housing loans. Figure 4.5 shows the effect of a one-standard-deviation shock to NEWLOAN on housing development since the third quarter of 1999, when the budget exceeded the actual value of GHLC-issued housing loans. According to Fig. 4.4, up to the second quarter of 1999, NEWLOAN increased new housing development between the third and the eighth quarter following the shock. Figure 4.5 shows that, since the third quarter of 1999, increases in NEWLOAN still contributed to a rise in new housing development between the third and sixth quarter following the shock.

4.3 The Benefits of Direct Housing Loans Issued by the GHLC

67

New Housing Development (Units)

20000 15000 10000 5000 0 -5000 -10000

1

2

3

4

5

6

7

8 Quarters

Fig. 4.4 Impulse response of new housing development to a shock in the value of housing loans of the GHLC (–the second quarter of 1999)

New Housing Development (Units) 10000 8000 6000 4000 2000 0 -2000

1

2

3

4

5

6

7

8

Quarters

Fig. 4.5 Impulse response of new housing development to a shock in the value of housing loans of the GHLC (the third quarter of 1999–)

68

4

Housing Loan Policy in Japan: Benefits and Drawbacks of Low-Interest. . .

Table 4.9 Impulse response of new housing development caused by a shock to the value of the GHLC housing loans

Quarters since shock 1 2 3 4 5 6 7 8

Impulse response (1std dev.) –1999Q2 1999Q3– 0.0 0.0 1966.9 1618.5 1806.0 1612.4 2510.9 1134.5 4878.3 758.2 8214.7 1438.3 7443.9 1787.0 6898.2 19.6

Accumulated impulse response (1std dev.) –1999Q2 1999Q3– 0.0 0.0 1966.9 1618.5 160.9 6.1 2350.0 1128.3 7228.3 1886.5 15,443.0 3324.8 22,886.9 1537.8 29,785.1 1518.3

Table 4.9 indicates the numerical value of impulse response and the accumulated impulse response of a one-standard-deviation shock to the residuals of NEWLOAN, which resulted in a rise in new housing development of 8214.7 units in the six quarters leading up to the second quarter of 1999. From the third quarter of 1999, an increment in NEWLOAN of one standard deviation induced an increase in new housing development of 1438.3 units in the six quarters. Next, I examine the accumulated impulse response. Up to the second quarter of 1999, a one-standard-deviation increase in the value of GHLC-issued housing loans resulted in an increase in new housing development of 29,785.1 units in eight quarters. Since the third quarter of 1999, an increase in NEWLOAN of one standard deviation raised new housing development by 1518.3 units in the eight quarters. It can therefore be concluded that the value of GHLC-issued housing loans did have an effect on new housing development. However, this effect weakened after the third quarter of 1999, when the budget exceeded the actual value of GHLC housing loans. As shown in Sect. 4.3.2.1, the demand for GHLC-issued housing loans exceeded the budget for such loans until the second quarter of 1999. The increase in the value of housing loans supplied by the GHLC raised housing loans granted along the demand curve and promoted more new housing development (as shown in Fig. 4.10 of Appendix 2, where the supply of housing loans moved from the vertical line, S1t, to S2t, NEWLOAN increased from q0 to q). I label this as the “direct effect.” As noted by Yoshino and Nakata (2000), credit rationing in the Japanese housing loan market during this period also produced an “indirect effect,” in that the increased value of NEWLOAN induced an increase in housing loans from private financial institutions, which also increased housing development. From the third quarter of 1999, when the budget exceeded the actual value of housing loans issued by the GHLC, credit rationing, and the resulting indirect effect, vanished. Thus, the impulse response of new housing development from the value of GHLC-issued housing loans decreased. However, the direct effect continued.

4.4 The Drawbacks of Direct Finance Under the GHLC

4.3.2.2

69

Variance Decomposition

A variance decomposition analysis is conducted to determine the importance of the interest rate on GHLC housing loans (RKOUKO) and of the value of GHLC-issued housing loans (NEWLOAN) to changes in new housing development. According to Table 4.10, up to the second quarter of 1999, 30.2% of the variation in housing development measured at eight quarters was caused by RKOUKO and 29.3% by NEWLOAN. Since the third quarter of 1999, the effect of RKOUKO declined sharply to 13.5%, while the impact of NEWLOAN remained at 32.1%. This result is consistent with the results of the impulse response functions.

4.4

The Drawbacks of Direct Finance Under the GHLC

In this section, to recognize the drawbacks of direct finance under GHLC, I firstly explain how interest rates were set within the housing loan market when the GHLC was issuing direct loans. Secondly, I indicate the problems caused by housing loans issued by the GHLC in relation to the oppression of commercial activities. Thirdly, using welfare analysis, I show the welfare loss of society as a whole caused by GHLC’s low-interest housing loans. Finally, I identify the negative impacts of GHLC-issued low-interest housing loans from the perspective of the general account’s deficit compensation.

4.4.1

Interest Rate

The interest rate of GHLC housing loans varied according to the interest rate of the FILP (Zaisei Tōyūshi)—the interest rate of loans issued by the Ministry of Finance to special corporations such as the GHLC. In addition, the interest rate for GHLC housing loans was set by the GHLC and not by the market. JHFA (2017) noted that

Table 4.10 Variance decomposition of new housing development (%)

Quarters 1 2 3 4 5 6 7 8

RKOUKO –1999Q2 4.3 6.3 10.6 15.7 22.1 28.0 30.2 30.4

1999Q3– 0.1 7.6 6.5 8.3 9.9 11.8 13.1 13.5

NEWLOAN –1999Q2 28.7 24.4 27.3 30.1 31.0 30.3 29.3 28.1

1999Q3– 20.3 20.5 27.0 28.3 28.7 30.8 32.1 31.8

70

4

Housing Loan Policy in Japan: Benefits and Drawbacks of Low-Interest. . .

the interest rates for housing loans under the GHLC were lower than those of private financial institutions up to the second quarter of 1999.

4.4.2

The Oppression of Commercial Activities

Research has indicated that GHLC-issued housing loans displaced the activities of private financial institutions (see, for example, Izu, 1999). Since the fixed interest rate of private financial institutions (commercial banks) was greater than that of the GHLC, consumers first attempted to borrow from the GHLC, and only borrowed the outstanding amount from private financial institutions if their applications were turned down or if their requirements were higher than that the GHLC was willing to supply.

4.4.3

Welfare Loss

In Fig. 4.6, Dt shows the demand for the GHLC’s housing loans. Since the GHLC issued loans at a fixed interest rate of RKOUKO, the supply curve of GHLC housing loans was then the horizontal S’ t rather than St (where St is the supply curve of GHLC housing loans when the GHLC issued loans under a non-fixed interest rate). Noting that GHLC issued housing loans at a lower interest rate than the market, I show that RKOUKO was lower than the market interest rate of rLt* (which would be the equilibrium market interest rate if GHLC housing loans were not issued at a fixed low rate). Excess demand in the housing loan market was present under the interest rate RKOUKO. In order to meet this excess demand, the GHLC increased the issuance of housing loans until q in excess of the budget. The equilibrium of demand and supply is marked as E0 . Consumer surplus increased by CDE0 E through the issuance of GHLC housing loans at low-interest rates, rather than equilibrium interest rates, where demand would meet the supply of GHLC housing loans at a non-fixed rate of interest. This indicates that consumers enjoyed a decrease in expenditure by virtue of the lower interest rate. If the GHLC were to issue housing loans at the equilibrium interest rate rLt*, its profit would be COqB, while if it issued housing loans at RKOUKO, its profit would be DOqE0 . Therefore, the GHLC-induced loss was CDE0 B. This opportunity cost was paid by the National Treasury. I subtract this loss from the rise in consumer surplus, then society’s net welfare loss from GHLC housing loans at a low-interest rate was ⊿EBE0 . In other words, this amount corresponds to the general account’s compensation of the deficit. This is the loss created by the GHLC housing loans because of low-interest rates, paid by taxpayers.

4.5 Conclusion

71

Fig. 4.6 Welfare loss from housing loans of the GHLC at low interest rates

As Fig. 4.6 shows, GHLC’s low-interest housing loans created a welfare loss to Japanese society as a whole because the taxpayers’ additional burden was greater than the increase in consumer surplus.

4.4.4

The General Account’s Compensation for the Deficit

As indicated in Sect. 4.2.2, the difference between the loan rates under the GHLC and the lending rate of the FILP was covered by subsidies from the government. According to Shiota et al. (1999), the tax revenues invested into the GHLC totaled ¥ 565.8 billion, which was the largest amount spent to special agencies and was a burden on taxpayers.

4.5

Conclusion

This analysis finds that both the low-interest rate policy and the value of the loans issued by the GHLC have contributed to the level of house building in Japan. Firstly, a lowering of the interest rate for GHLC housing loans by one standard deviation of the residuals increased new housing development by 6770.7 units in five quarters. This effect had disappeared in the third quarter of 1999, when the GHLC budget exceeded the actual value of housing loans issued. Secondly, a one-standard-deviation increase in the value of GHLC housing loans raised housing construction by 8214.7 units in the 15 months up to the second quarter of 1999. In the third quarter of 1999, this effect had diminished by around 80% but still existed.

72

4

Housing Loan Policy in Japan: Benefits and Drawbacks of Low-Interest. . .

These results identify the contribution of GHLC policy to the size of the housing stock in Japan. The GHLC also issued loans to provide barrier-free, senior-friendly, and earthquake-proof homes that improved the quality of housing (GHLC, 2003). On the other hand, GHLC housing loans suppressed the activities of private financial institutions and caused a budget deficit, thereby imposing a burden on taxpayers, while also causing a dead weight loss through lower-than-market interest rates. Moreover, in reality, housing loans supported by securitization under the JHFA accounted for only 10% of all housing loans, since they were difficult for private financial institutions to manage. In addition, the high costs made issuing housing loans unprofitable for private financial institutions in an environment of low interest rates. Considering these outcomes, it could be claimed that Japan’s housing loan policy was beneficial despite the drawbacks. However, it is difficult to imagine a future revival of the GHLC direct loan system due to a lack of political support. The following chapters build upon this analysis by asking the following questions: (1) If the role that the GHLC played in the past is still necessary today, would administrative corporations be able to provide these services? and (2) Is it possible for an administrative corporation to implement policies compliant with present needs and improve the quality of housing? Calculating the dead weight loss by estimation of the demand curve and the supply curve of the housing loan market, which shows welfare loss caused by the GHLC housing loans numerically, is left for our future study.

Appendix 1: Data Used for Estimation (1) The interest rate on GHLC housing loans The GHLC standard interest rate (kijunkinri) from the GHLC (2) The value of GHLC housing loans granted The value of new individual housing loans granted by the GHLC, Financial and Economic Statistics Monthly, the Bank of Japan (3) New housing development New individual housing development (units, the sum of new housing development owned and built for sale, i.e., without new housing development rented and issued), Monthly Construction Statistics, Ministry of Land Infrastructure and Transport.

Appendix 2: Seasonality of Variables in the VAR Models In order to check seasonality of my data, I plot the data in Figs. 4.7, 4.8, and 4.9.

Appendix 2: Seasonality of Variables in the VAR Models

73

6 5 4 %

3 2 1 80 81 82 83 84 85 86 87 88 89 90 91 92 93 94 95 96 97 98 99 00 01 02 03 04 Year

Fig. 4.7 Interest rate of GHLC housing loans

35,000

Billion Yen

30,000 25,000 20,000 15,000 10,000 5,000 0

80 81 82 83 84 85 86 87 88 89 90 91 92 93 94 95 96 97 98 99 00 01 02 03 04 Year

Fig. 4.8 Amount of the GHLC housing loans granted

280,000 260,000

Units

240,000 220,000 200,000 180,000 160,000 140,000

80 81 82 83 84 85 86 87 88 89 90 91 92 93 94 95 96 97 98 99 00 01 02 03 04 Year

Fig. 4.9 New housing development in Japan

74

4

Housing Loan Policy in Japan: Benefits and Drawbacks of Low-Interest. . .

Demand for housing loan of the GHLC Supply of housing loan of the GHLC

Fig. 4.10 Housing loan market with low-interest rate housing loan of the GHLC

Appendix 3: Housing Market When the Actual Value of Housing Loans of the GHLC Exceeded the Budget Here we look at housing market when the actual value of housing loans of the GHLC exceeded the budget. This occurred in the period up to the second quarter of 1999. In Fig. 4.10, Dt is demand curve of housing loans of the GHLC. The GHLC made loans at a fixed interest rate RKOUKO. We show the budget of GHLC housing loans by q0 . Reflecting that this is the analysis about the period when the actual value of housing loans of the GHLC exceeded the budget, RKOUKO was lower than equilibrium market interest rate rLt*, which means that there was excess demand for housing loans at interest rate RKOUKO. Facing this excess demand, the GHLC made housing loans (thus the actual value) over the budget till q. The supply curve of the GHLC housing loans shifted from S1t to S2t.

References GHLC (1993) The Japanese Housing Loan, Tokyo: Housing Loan Progress Association. GHLC (2003) Housing Loan Corporation Yearbook, Tokyo: GHLC. Hirono, K. N. (1998) “Estimation of Housing Price Index and Rent Index,” Josai University Bulletin, 16(1), 69–78. Hirono, K. N. (2005) “A VAR Analysis of the Effect of Housing Loan Policy,” Pacific Economic Review, 10(4), 557–76. Hirono, K. N. (2020) “Low-interest Rate Policy and Japanese Housing Market,” Keizai-shushi (The Nihon University Economic Review), 90(1), 41–57.

References

75

Homma, M., N. Atoda, K. Fukuma and T. Asada (1988) “Housing Policy and Housing Demand,” Financial Review, Ministry of Finance, 7, 4–20. Housing Loan Progress Association (2018) The Pocket Housing Statistics, Tolyo: Housing Loan Progress Association. Izu, H. (1999) Changing Housing Market and Housing Policy, Tokyo: Toyokeizaishinpousha. JHFA (2017) Research on Lending Trend of Individual Housing Loan in 2017 Fiscal Year, Tokyo: JHFA. Kamoike, O. (1991) “The effects of Housing Finance Policy,” Jutakutochikeizai (Quarterly Journal of Housing and Land Economics), 2, 20–24, Moriizumi, Y. (1996) “Credit Rationing and Public Housing Loans in Japan,” Journal of Housing Economics, 5, 227–46. Pozdena, R. J. (1990) “Do Interest Rates Still Affect Housing?” Federal Reserve Bank San Francisco Economic Review, 3–14. Pyndick, R. S. (1981) Econometric Models and Economic Forecasts, Singapore: McGraw-Hill. Shiota, H., T. Hiromatsu, M. Yamaguchi and S. Kawakami (1999) “Ranking of Degree of Failure of Special Corporations,” Nikkei Business, September, 22–26. Wago, H. (1987) “Time Series Analysis of Public and Financial Policy,” in Yabushita, S. and K. Asako (eds) Japanese Economy and Public Policy: Analysis of Macroeconomy and Budget Deficit, Tokyo: Toyokeizaishinposha. Yamasawa, N. (2004) Introduction to Practical Econometrics, Tokyo: Nihonhyoronsha. Yamasawa, N. and K. Nakano (1998) “Stability of VAR Model and Error Correction Model: Empirical Study on Fiscal Policy Impacts,” Japan Center for Economic Research, 91. Yasui, R. (2015) “Review of History of Japanese Housing Loan: the Period of Maturity of Housing Loan (2007–),” Evaluation, 56, 74–87. Yoshino, N. and M. Nakata (2000) “Reformation of the Government’s Investment and Loan Program and Future Public Housing Finance,” Jutakutochikeizai (Quarterly Journal of Houssing and Land Economics), 38, 20–27.

Chapter 5

Implementation of Subsidy for Improving the Earthquake-Proof Conversion of Rental Housing

Abstract A house that is not earthquake-proof suffers from external diseconomies. For example, the collapse of such houses causes diseconomies in the form of disrupted firefighting efforts and casualties. Thus, it is crucial to formulate housing policies to make rental houses more earthquake-proof. In the context of Japan, this study explores the cause of the low rate of earthquake-proof conversions of rental houses and proposes a method to estimate the amount of subsidy needed to improve the situation. Hence, the study explains the great earthquakes in Japan, mainly the Great East Japan Earthquake. Subsequently, it develops a method to examine consumers’ evaluation of the earthquake-proof conversions of rental houses and determines a rental firm’s profits from such conversions. I apply these methods to the rental apartments in Tokyo and the Miyagi prefecture. I find that tenants are willing to pay less than 3000 yen/month in addition to the present rent, for the earthquakeproof conversion of houses. This finding reveals the unprofitability of the earthquake-proof conversion of houses for the rental firms and the need for subsidies to make rental houses more earthquake-proof in Japan. Based on the analysis, this study proposes a method to determine the amount of subsidy needed for the earthquake-proof conversion of rental houses. Specifically, it provides a framework for municipal subsidies for such conversion. This framework has policy implications for government subsidies for earthquake-proofing houses. The findings have implications for raising awareness for the earthquake-proof conversion of houses and thereby increasing tenants’ willingness to pay for and, in turn, the rental firms’ profits from such construction. Keywords Earthquake · Housing policy · CVM · Willingness to pay · Subsidy · Earthquake-proof houses

5.1

Introduction

In 2011, the Great East Japan Earthquake completely and partially destroyed 127,800 and 275,807 houses, respectively, and claimed enormous number of precious lives (Taniguchi, 2016). Given the occurrence of such and frequent © Springer Nature Singapore Pte Ltd. 2022, corrected publication 2022 K. N. Hirono, Economic Analysis of Housing Policy in Japan, New Frontiers in Regional Science: Asian Perspectives 64, https://doi.org/10.1007/978-981-19-4925-8_5

77

78

5 Implementation of Subsidy for Improving the Earthquake-Proof Conversion. . .

earthquakes in Japan, it is necessary to earthquake-proof houses to reduce the impact of seismic risk, and thereby alleviate human and material losses. A house that is not earthquake-proof suffers from external diseconomies. For example, the collapse of such houses causes diseconomies in the form of disrupted firefighting efforts and casualties. Thus, it is crucial to formulate housing policies to make rental houses more earthquake-proof. According to a survey by Sato (2018), 20.9% of the houses in Japan are not earthquake-proof, and the implementation rate of the earthquakeproof conversions is only 3.5%. In the given context, in this chapter, firstly, I develop a method to examine consumers’ evaluation of the earthquake-proof conversions of rental housing. Secondly, I present a method to determine a rental firm’s profits from construction work to make rental houses earthquake-proof. Thirdly, based on the above analysis, I provide a method to determine the amount of the subsidy needed for the earthquakeproof conversion of rental housing. Fourthly, using the data on rental apartments in Tokyo and the Miyagi Prefecture, I estimate the consumers’ willingness to pay for the earthquake-proof conversions of rental houses and the factors influencing this willingness to pay. I also estimate the rental firm’s profits from and the necessary amount of subsidy for the earthquake-proof conversions of rental houses. Concerning the rental market, I deal with a monopolistic market of rental housing, where the rental houses are differentiated by earthquake-proof qualities (attributes). Firms enter this rental house market—the market of earthquake-proof rental housing—if they earn profits from making rental houses earthquake-proof. In the next section, I explain the major earthquakes that have taken place in Japan, mainly the Great East Japan Earthquake, to introduce this study’s background and aim, and to facilitate an understanding of this chapter. In this chapter, I analyze the earthquake-proof conversion of apartments for the following reasons. First, less than 30% of the local governments subsidize the earthquake-proof conversion of apartments. Second, the ratio of non-earthquake-proof apartments accounts for 17.01% and 22.26% of the wooden and non-wooden apartments, respectively (calculated by the author using the data of the Statistic Bureau, Ministry of Internal Affairs and Communications, 2017). This finding implies a high number of non-earthquakeproof apartments. Third, apartments are homogeneous, relative to detached houses, which ensures the accuracy of the outcome of this chapter. The contingent valuation method used in this chapter requires homogeneity in the attributes of the houses analyzed. The lack of homogeneity would lead to partial– overall bias, given that tenants’ willingness to pay would vary if there is a variance in their perception of the houses’ qualities. Specifically, relative to apartments, detached houses vary by layout, which leads to variance in consumers’ perception of houses. This variance will lead to a difference in consumers’ willingness to pay if they are asked about the willingness to pay for the earthquake-proof conversion of detached houses. Thus, an analysis of detached houses would lead to partial–overall bias, and thereby lower the accuracy of the outcome. As stated earlier, this study focuses on the earthquake-proof conversion of rental apartments. This has the following reasons. Firstly, firms that rent houses will carry out the earthquake-proof conversion projects if they generate positive net profits

5.1 Introduction

79

from the project. In this context, it must be noted that the availability of subsidies can cover the differences between the project’s profit and cost, thereby advance the project. Conversely, individually owned houses are not earthquake-proofed if these owners fail to recognize the need for earthquake-proof housing, lack time to execute the conversion (including the time to consult the constructor), lack information about seismic reinforcement and the right contractor in addition to financial matter. Secondly, the non-earthquake-proof houses suffer from external diseconomies, which can be efficiently resolved by utilizing the expertise of housing companies. Thus, I analyze rental housing managed by the housing companies. It must be noted that earthquake-proof conversion of rental apartment is not progressing smoothly. This delay can be attributed to the fact that the rental companies fail to get the expected returns from their earthquake-proof conversion projects. In this context, I estimate consumers’ willingness to pay for the earthquakeproof conversion of rental apartments and I provide a method to determine whether it is profitable for companies to make these conversions. I also propose a method to assess if and how much subsidy is required for the earthquake-proof conversion of rental apartments. I apply these methods to the rental apartments in the Tokyo metropolitan area and the Miyagi Prefecture. Specifically, using the contingent valuation method, I calculate how much additional rent consumers pay when companies earthquake-proof their rental apartments. Using the outcome of this estimation and the discounted cash flow (DCF) method, I determine whether the rental companies make profits from the earthquake-proof conversion of rental apartments. If they do not earn enough profits to cover the cost, then, based on the rental firms’ net present value, the local governments can implement the method proposed in this chapter to calculate the subsidy required to accelerate the earthquake-proof conversion of rental apartments. Specifically, I calculate the amount of subsidy needed for the rental apartments in Tokyo and Miyagi.1 Alonzo (2002) uses the contingent valuation method (CVM) to estimate the consumers’ willingness to pay for the benefits of barrier-free housing in Madrid and Barcelona. Sato et al. (2005) use the contingent valuation method to analyze the demand and willingness to pay for barrier-free housing in the Tokushima Prefecture. Sato and Tamamura (2006) use the contingent valuation method to evaluate the earthquake-proof conversion of houses. The differences between Sato and Tamamura (2006) and this chapter are as follows. Firstly, Sato and Tamamura (2006) mainly analyze owner-occupied detached houses, whereas this chapter investigates rental apartments. This chapter also analyzes the disaster area, which provides a context for using the experiences from the Great East Japan Earthquake. Secondly, Sato and Tamamura (2006) do not ask their survey participants about the level of earthquake-proofing of houses in the question on the willingness to pay for the conversion. Since the willingness to pay for the earthquake-proof conversion

1

Refer to Hirono (2019) about the CVM and DCF methods in general. Hirono (2019) presents the method to calculate consumers’ evaluation of the earthquake-proof conversion of housing. Data and a part of the analysis in this chapter are from Hirono (2020a, b).

80

5 Implementation of Subsidy for Improving the Earthquake-Proof Conversion. . .

depends on the level of earthquake-proofing in the house, I define the level of earthquake-proof conversion as the level that helps a building to withstand an earthquake with a seismic intensity of seven on the Japanese seven-stage seismic scale. Thirdly, this chapter not only calculates the consumers’ willingness to pay for the earthquake-proof conversion of rental apartments using the contingent valuation method but also estimates whether companies can earn profits from such conversions combining the DCF method with CVM. Yamaga et al. (2002) present pioneering research concerning the earthquakeproof conversion of houses. Estimating the hedonic function of rents, these authors show the difference of rents between rental apartments in accordance with the earthquake-resistant design standards set after 1981 and the apartments constructed based on the seismic code before 1981 in Tokyo. They also determine the profitability of the earthquake-proof conversion of houses. This research differs from that of Yamaga et al. (2002) in the following aspects. Firstly, this research does not adopt the earthquake-resistant design standards set after 1981 because houses conforming to this standard have been damaged by earthquakes with a seismic intensity of 7. Secondly, this study considers not only Tokyo but also the Miyagi Prefecture, which is one of the disaster areas of the Great East Japan Earthquake. Thirdly, while Yamaga et al. (2002) calculate the increment in rents from the earthquake-proof conversion using the hedonic rent function, this research uses the contingent valuation method. This can be attributed to the lack of data showing whether each rental house can withstand an earthquake with a seismic intensity of 7, which is needed for estimating rents using the hedonic rent function. Fourthly, this research newly proposes a method to calculate the subsidy required to support the earthquake-proof conversion of rental apartments. Moreover, this research first discusses factors influencing the consumers’ willingness to pay for the earthquake-proof conversion of rental apartments. Finally, I propose a system to adjust the amount of subsidy according to consumers’ attributes. Briefly, this is the first research to analyze the following points. First, it estimates the consumers’ willingness to pay for the earthquake-proof conversion of rental apartments to withstand an earthquake with a seismic intensity of 7 using the contingent valuation method. This willingness to pay is consumers’ evaluation of such a conversion. Second, the research determines the profits of the rental firms from earthquake-resistance construction of rental apartments in the Tokyo metropolitan area and the Miyagi Prefecture, using the DCF method. Third, this research constructs a framework to determine the need for and amount of subsidy. Fourth, it explores the reasons for variance in the consumers’ willingness to pay for the earthquake-proof conversion of rental apartments. The rest of the chapter is organized as follows. Section 5.2 explains the major earthquakes that have occurred in Japan. Section 5.3 presents the data and method used for analysis. Section 5.4 reports and interprets the outcome of the analysis. Section 5.5 identifies the reasons for variance in the consumers’ willingness to pay for the earthquake-proof conversion of rental apartments. Section 5.6 proposes a method to calculate the amount of subsidy required to accelerate the

5.2 Earthquakes in Japan

81

earthquake-proof conversion of rental housing and a method to formulate a system of the subsidy. Section 5.7 presents the conclusions.

5.2

Earthquakes in Japan

This section emphasizes the Great East Japan Earthquake, which motivated the research in this chapter, and discusses the purpose of this research. The Great East Japan Earthquake of magnitude 9.0 was caused by a massive earthquake off the coast of Tohoku on March 11, 2011. With a strong quake of a seismic intensity of 7 and associated tsunamis, the Great East Japan Earthquake brought catastrophe to the east of Japan and caused a meltdown in Tokyo Electric Power’s Fukushima I Nuclear Power Plant. Concerning the human damage, the earthquake claimed 15,899 lives and injured 6157 people (National Police Agency, 2020). These casualties were mainly in the Miyagi, Iwate, and Fukushima Prefectures (National Police Agency, 2011). As for the material damage, the earthquake destroyed an enormous number of houses completely or partially. In this context, it must be noted that the earthquakes in Japan are categorized into interplate and intraplate earthquakes. While the former occurs when an oceanic plate pulls a continental plate underground causing the continental plate to leap up, the latter is caused by stresses within the plate. It must be noted that Japan is surrounded by many plates such as the North American, Pacific, Philippine Sea, and the Eurasia plates; among these plates, the oceanic plates are sinking under the continental plates. The Japan Meteorological Agency (2021) forecasts that this topography of Japan can contribute toward several major earthquakes in the future. In the past, Japan has witnessed several earthquakes with a magnitude of 7 or above. Some such earthquakes are the Kumamoto earthquake (2016), the Tokachi-Oki earthquake (2003), the Great Hanshin-Awaji earthquake (1995), the South China Sea earthquake (1946), the Tōnankai earthquake (1944), the Sanriku earthquake (1933), the Great Kanto earthquake (1923), and the Sanriku earthquake (1896). Despite the several earthquakes in the past, around 20% of the houses in Japan lack earthquake resistance (Sato, 2018). The lack of earthquake resistance poses a risk to the lives of Japanese people. This chapter aims to reduce such a risk. This chapter is a part of the research that I aim to release in countries affected by and vulnerable to earthquakes such as New Zealand, the United States (California and Alaska), China, Indonesia, Chile, the Philippines, India, and Turkey.

82

5.3 5.3.1

5 Implementation of Subsidy for Improving the Earthquake-Proof Conversion. . .

Method of Analysis and Data Discounted Cash Flow Method

This chapter adopts the DCF method as an investment criterion for the earthquakeproof conversion of houses. According to the DCF method, if the net present value (NPV) is positive, then the earthquake-proof conversion of rental housing will be profitable, and if the NPV is negative, then it will not be profitable. The NPV equals to the difference between the discounted present value of the present and future cash flows minus the cost of the earthquake-proof conversion of rental houses, where the cash flows are the increments in the rent created by the conversion. The equations are expressed as follows. I assume that a company renting a house conducts the earthquake-proof conversion of rental housing. Denoting increment in the annual rent by X yen, the discount rate by r%, and the remaining useful life of the house by n, the discounted present value Q can be stated as Q ¼ X=ð1 þ r Þ þ X=ð1 þ r Þ2 þ ⋯ þ X=ð1 þ r Þn :

ð5:1Þ

Since the NPV equals to the discounted present value of the present and future increments in the rent minus the cost, the NPV of the earthquake-proof conversion of this rental house is NPV ¼ X=ð1 þ r Þ þ X=ð1 þ r Þ2 þ ⋯ þ X=ð1 þ r Þn  C,

ð5:2Þ

where C is the cost of the earthquake-proof conversions. I assume that Xs remain constant for all the periods. I assume that the rental housing market is characterized by monopolistic competition. The rental firms conduct the earthquake-proof conversion of rental houses and enter the market if the NPV is positive, that is, when the earthquake-proof conversion is profitable.

5.3.2

Contingent Valuation Method

This chapter applies the contingent valuation method (CVM) to estimate the additional rent paid by consumers for the earthquake-proof conversion of rental apartments. Specifically, I use the CVM to calculate X in Eq. (5.2), which presents the formula to calculate the NPV of the earthquake-proof conversion of a rental apartment. The CVM is employed to ask consumers their willingness to pay for an improvement in the level of goods not traded in the market (non-market goods), after explaining the consumers the contents of that good (Hidano, 1999).

5.3 Method of Analysis and Data

83

Ciriacy-Wantrup (1947) introduces the concept of CVM, in the United States, in 1947. Subsequently, in 1963, in an analysis, Davis (1963) pioneers the application of the CVM to the value of outdoor recreation in woods (Kuriyama, 1998; Hidano, 1999). In the 1980s, the application of CVM increased in the studies of the other countries. Concerning the theoretical analysis of CVM, Hanemann (1984) introduces a random utility model, and Cameron (1988) proposes a model based on the willingness to pay. Duffield and Patterson (1991) present a non-parametric approach without assuming a distribution function. With these theoretical backgrounds, CVM became the most used method in the evaluation of non-market goods. The CVM has application in a wide range of analysis, and not only for evaluating the ecosystem or the value of tropical forest. The scope of CVM’s application has become wider. It has been used in recent empirical studies in the fields of medicine (Taylor et al. 2010; Yokochi et al. 2014), criminology (Cohen et al. 2004), stadium construction (Ruszkowski, 2017), and autonomous driving functions (Morita and Managi, 2018). These applications of CVM are aimed at determining the value of non-traded products. Concerning the field of construction, Alonzo (2002) applies CVM to barrier-free housing. In this context, it must be noted that the conditions for market existence are price and the volume demanded by the consumers; however, it is difficult to observe the price of being barrier-free. In this case, the price is determined by how much more consumers pay for barrier-free than non-barrier-free housing. Since this additional price cannot be observed, there is no market for the barrier-free quality of a house (Alonzo, 2002). Hence, Alonzo (2002) applies CVM to determine the pricing for barrier-free housing. Similarly, there is no market for the earthquake-proof conversion of rental houses. In most cases, the advertising catalogs and listings in housing advertisement magazines fail to provide information about whether a rental house is earthquake-proof. This keeps consumers from knowing the rental price of the earthquake-proofed rental houses—the difference between the rental price of an earthquake-proofed house and that of a non-earthquake-proofed house. Since this rental price cannot be observed or captured by consumers, there is no market for the earthquake-proof conversion of rental house. Although price has aspects of an evaluation of the goods and services, rental companies cannot grasp evaluation of the earthquake-proof conversion and its price, that is, an increment in the rent for the earthquake-proof conversion of the rental house. Therefore, it is appropriate to use CVM to estimate the rent appraisal for the earthquake-proofed rental houses.

5.3.3

Procedure of Estimation

In this research, I calculate X in Eqs. (5.1) and (5.2), by adopting CVM. I use the data on rental apartments in the Tokyo metropolitan area and the Miyagi Prefecture. The Miyagi Prefecture includes the disaster area of an earthquake with a seismic intensity of 7. Specifically, by using the questionnaire survey in Sect. 5.3.4, I ask the consumers their willingness to pay for the earthquake-proof conversion of rental

84

5 Implementation of Subsidy for Improving the Earthquake-Proof Conversion. . .

apartments. Subsequently, by applying the DCF method, I estimate the NPV of the future increment in rent due to the earthquake-proof conversion. If the NPV has a positive sign, then the rental companies will carry out the construction work. Conversely, if the NPV is negative, there would be a need for governments’ subsidy to accelerate the conversion. Given this, if the NPV has a negative sign, then the subsidy amount should be at least the absolute value of the NPV.

5.3.4

Questionnaire Survey on the Earthquake-Proof Conversion of Rental Apartments

1. Sampled period and respondents The questionnaire survey on the earthquake-proof conversion of rental apartments covers the period from February 3–7, 2017. The respondents of the questionnaire are the tenants of the apartments in Tokyo and the Miyagi Prefecture, whose commuting time from the center of the prefectures (Tokyo and Sendai stations, respectively) is less than 1.5 h. I restrict the sample to respondents with a personal income and aged between 20 and 70 years.2 This is because the survey takes the payment of rental by the respondents into account. I analyze the Miyagi Prefecture because it is the only prefecture hit by an earthquake of a seismic intensity of 7 in the Great East Japan Earthquake. I also analyze Tokyo because consumers’ attitude toward the earthquake-proof conversion may depend on whether the prefecture is in the disaster-stricken area. Another reason for choosing Tokyo is the high income of residents in Tokyo, relative to those in the Miyagi Prefecture, given that the income of residents may influence their willingness to pay for the earthquake-proof conversion of rental houses. Here, I assume the model of a monocentric city. In other words, I presume that the housing market expands around the center of the city—the Tokyo and Sendai stations for the Tokyo and Miyagi Prefectures, respectively. According to Hirono (2004), the region requiring residents to take more than 90 min of commute to the city’s center constitutes a different housing market. The consumers of this market emphasize different attributes of rental houses (from the attributes which consumers whose commuting time of less than 1.5 h to the center of the city think most important when choosing a rental home). Therefore, I choose respondents whose commuting time to the center of the city is less than 1.5 h. The survey is carried out online by a research company. Although an online questionnaire survey may be disadvantageous in collecting responses from respondents aged over 70 years, the survey helped in collecting sufficient correct responses from respondents of this age group.

2

The respondents in the seventies age group earn a pension which is a kind of personal income. Thus, I include this age group in the questionnaire.

5.3 Method of Analysis and Data

85

As this chapter is based on the online questionnaire survey, the respondents were registered as questionnaire monitors in the research company. The sampling method is purposive sampling. The population of this survey is those who live in rental housing in Tokyo and Miyagi Prefecture. The questionnaire survey has 186 active parts (100 and 86 in Tokyo and the Miyagi Prefecture, respectively). I manage the sampling in such a manner that the proportion of the elderly and household income match those of the population, given that these aspects seemed to have relevance to the willingness to pay for the earthquake-proof conversion of rental housing. In general, there may emerge a bias toward online survey if it involves the elderly samples, given that it is difficult to collect samples of the elderly. Thus, I match the proportion of the elderly with that of the population. For non-random selection, I set the proportion of respondents above 60 years to 30%, which is same as the proportion in the rental housing data collected by the Ministry of Internal Affairs and Communications (2017). Concerning the household income, I set the distribution ratio of sampling according to the Ministry of Land, Infrastructure, Transport and Tourism (MLIT) (2017). For example, the tenants with a household income of 4–6 million yen account for 27.4% of the respondents in this data and 26.9% in MLIT (2017). The proportion of tenants with a household income of 6–8 million yen represents 14.5% of the respondents in this data and 13.3% in MLIT (2017). The proportion of tenants earning 8–10 million yen represents 6.5% in this data and 5.7% in MLIT (2017). 2. Content of the Questionnaire First, I collect the demographic details of the respondents—age, gender, household income, and the place of residence. I also enquire about the type of house in which they live—rental or owned house, wooden or non-wooden apartment, or a detached house—and their experience of an earthquake with a seismic intensity of above 6 in the previous 10 years. Second, I ask them the need of earthquake-proof conversion of rental houses to withstand an earthquake with a seismic intensity of 7. This question intended to gauge respondents’ awareness of the need for the earthquake-proof conversion of rental homes. Finally, I ask about the consumers’ willingness to pay for the earthquake-proof conversion of rental houses. 3. Attributes of the respondents Among the respondents, male and female respondents comprise 50.54% (94 people) and 49.46% (92 people), respectively. The highest proportion of respondents is in the thirties, accounting for 27.42% (51 people), followed by 26.34% (49 people) respondents in the sixties (Fig. 5.1). Concerning the household income, the proportion of households earning 2–4 million yen is the largest at 37.63% (70 people), followed by 27.42% (51 people) households with an income of 4–6 million yen (Fig. 5.2). The respondents with an experience of an earthquake with a seismic intensity of above 6, in the previous 10 years, represent 50% (93 people) of the respondents.

86

5 Implementation of Subsidy for Improving the Earthquake-Proof Conversion. . . 5%

10%

26% 28%

10% 21% twenes

thires

fores

fiies

sixes

sevenes

Fig. 5.1 Age of the respondents Fig. 5.2 Household income of the respondents

2% 3%

9%

6% 15% 38% 27%

Less than 2 million yen

2-4 million yen

4-6 million yen

6-8 million yen

8-10 million yen

10-12 million yen

More than 12 million yen

4. Questions on the willingness to pay I formulate four types of questions on the willingness to pay for the earthquake-proof conversion of rental houses, as shown in Fig. 5.3. I ask the residents of Tokyo about wooden (Question 1) and non-wooden (Question 2) apartments. I ask the residents of the Miyagi Prefecture the questions about wooden (Question 3) and non-wooden apartments (Question 4), respectively. The choices of the willingness to pay are 0, 1–499, 500–999, 1000–1499, 1500–1999, 2000–2499, 2500–2999, 3000–3499, 3500–3999, 4000–4499, 4500–4999, 5000–5499, 5500–5999, 6000–6499, 6500–6999, 7000–7499,

5.3 Method of Analysis and Data

87

(Question 1: A question to the respondents in the Tokyo Metropolitan area about evaluation of earthquake-proof conversion of wooden rental apartment) How much are you willing to pay additional monthly rent for earthquake-proof conversion of wooden rental apartment

an earthquake with a seismic intensity of seven. Rent is

63,000 yen now, the area is 27 m 2, the age is 23 years, the commuting time to the Tokyo station is 39 minutes about this apartment. (Question 2: A question to the respondents in the Tokyo Metropolitan area about evaluation of earthquake-proof conversion of non- wooden rental apartment) How much are you willing to pay additional monthly rent for earthquake-proof conversion of nonwooden (Steel-reinforced concrete structure or steel structure) rental apartment to withstand an earthquake with a seismic intensity of seven. Rent is 90,000 yen now, the area is 37 m 2, the age is 18 years, the commuting time to the Tokyo station is 39 minutes about this apartment. (Question 3: A question to the respondents in the Miyagi Prefecture about evaluation of earthquakeproof conversion of wooden rental apartment) How much are you willing to pay additional monthly rent for earthquake-proof conversion of wooden rental apartment

an earthquake with a seismic intensity of seven. Rent is 2

46,000 yen now, the area is 35 m , the age is 17 years, the commuting time to the Sendai station is 24 minutes about this apartment. (Question 4: A question to the respondents in the Miyagi Prefecture about evaluation of earthquakeproof conversion of non- wooden rental apartment) How much are you willing to pay additional monthly rent for earthquake-proof conversion of nonwooden (Steel-reinforced concrete structure or steel structure) rental apartment

an

earthquake with a seismic intensity of seven. Rent is 55,000 yen now, the area is 39 m 2, the age is 17 years, the commuting time to the Sendai station is 24 minutes about this apartment.

Fig. 5.3 Questions about evaluation of earthquake-proof conversion of rental apartment

7500–7999, 8000–8499, 8500–8999, 9000–9499, 9500–9999 yen, more than 10,000 yen. I ask the respondents to choose one of these ranges. Specifically, I ask the respondents the additional amount of rent they will pay for the earthquake-proof conversions of rental apartments. I present the choices in the units of 500 yen because of the following reasons. Forento (a Japanese rental housing magazine) lists most of the rents in the units of 1000 or 10,000 yen. In the Japanese rental housing market, rents are set in the units of 1000 or 10,000 yen, and consumers’ expectation of rents corresponds to these units. However, the rents of a few rental houses are listed in the units of 500 yen such as 95,500 yen. Thus, concerning the

88

5 Implementation of Subsidy for Improving the Earthquake-Proof Conversion. . .

choices provided in the questionnaire, the price range is set narrowly using the units of 500 yen to avoid leakage. Let X be the average value of the willingness to pay for the earthquake-proof conversions of rental apartments (i.e., answer to questions in Fig. 5.3). Using this X, I calculate the NPV for this conversion in Eq. (5.2). Concerning the area, the building’s age, and the commuting time, in Questions 1–4 in Fig. 5.3, I use the average value of those (the area, the building’s age, and the commuting time) of the wooden and non-wooden private rental apartments in Tokyo and the Miyagi Prefecture, respectively, from the data provided by the Ministry of Internal Affairs and Communications (MIC) (2017). As for the rents in Fig. 5.3, I use the monthly rent data provided by MIC (2017) for wooden and non-wooden private apartments in Tokyo and the Miyagi Prefecture. Concerning the earthquake-proof level, some studies determine whether the houses are earthquake-proof based on the new seismic code for houses set in 1981. However, I define the earthquake-resisting performance of houses based on whether houses can resist earthquakes with a seismic intensity of 7. Therefore, I ask about the willingness to pay rent for conversions that help rental apartments resist earthquakes with a seismic intensity of 7. This is because houses built based on the new seismic code can resist only earthquakes with a seismic intensity of 6; thus, the new seismic code falls short as a seismic code. For example, the Great HanshinAwaji Earthquake with a seismic intensity of 7 caused complete and medium or small damage to 8.6% and 16.7%, respectively, of the buildings conforming to the new seismic code (Yoshizawa Architectural Structure Design, 2013). Thus, I define the earthquake-resistant houses as houses that can resist earthquakes with a seismic intensity of 7, which is the highest earthquake scale. I set the quality of rental houses in terms of the area, building age, commuting time to the city center, and the type of apartment (wooden or non-wooden house), in Questions 1–4 in Fig. 5.3.3 This approach allows the respondents to capture the concrete image of the rental house in the question. I ask the respondents their willingness to pay for these rental houses with the specified quality, because the CVM is the method to ask the consumers their willingness to pay for goods or services of a certain quality. For instance, if I use the general word “earthquakeproof conversion of house” in the question, which includes all the houses, such as apartment, detached house, wooden and non-wooden house, then there would be a significant variance in the respondents’ willingness to pay.

3

The prices of the apartments depend only on the area, the age of the apartments, and the commuting time to the city center, if construction of the apartments is same (Hirono, 2004). Thus, in the question on the willingness to pay, I set these as the quality parameters of a house.

5.4 Outcome of Analysis

5.3.5

89

Data

In this chapter, I determine the profitability of the earthquake-proof conversion of rental housing by estimating Eq. (5.2). In this procedure, as the discount rate, I use the real interest rate on 10-year government bonds.4 To calculate Eq. (5.2), I set the lifespan of housing as 32.2 years, based on the research by MLIT (1996). The remaining useful life of the house n—the period for which the future additional rents for earthquake-proof conversion is discounted—is calculated by the lifespan of housing minus the average age of housing. I calculated this period as for the wooden and non-wooden rental apartments in Tokyo and the Miyagi Prefecture, respectively. I set the cost of making the rental houses earthquake-proof at 50,000 yen/m2, which is the average of the data in the earthquake countermeasure website called Taishin Net (2017).5 I calculate the cost of earthquake-proof conversion of rental houses in Tokyo and Miyagi by multiplying rental apartments’ average area (in each prefecture) and 50,000 yen. The data used in this chapter are the respondents’ willingness to pay for the earthquake-proof conversion of rental housing X and the NPV.

5.4 5.4.1

Outcome of Analysis Awareness of the Earthquake-Proof Conversion of Housing

Firstly, based on the outcome of the questionnaire survey, I present consumers’ perception of the need of earthquake-proof conversion of rental housing. According to Fig. 5.4, 84% of the consumers were aware of the need for making rental houses resistant to earthquakes with a seismic intensity of 7. Overall, consumers strongly felt the need for earthquake-proof rental housing.

4 By denoting the inflation rate as π and nominal interest rate m, the value of 1 yen will be 1 + m  π yen, that is, the value of money will decrease by inflation. Hence, I set the real interest rate as the discount rate. I use the real interest rate on 10-year government bonds as the discount rate, because the long-term investment is required for earthquake-proofing rental houses. 5 The tenant-occupied rental houses can be made earthquake-proof using the following methods: fitting steel frame into the walls of the house from the corridor, reinforcing using a strong structure, and out framing (fitting steel frame from outside of the building) (New Japan Reform, 2019).

5 Implementation of Subsidy for Improving the Earthquake-Proof Conversion. . .

90

Fig. 5.4 The need of earthquake-proof conversion of rental housing

16%

84% Yes

No

Table 5.1 The willingness to pay for construction work to make rental houses earthquake-proof (Yen/month) Wooden Non-wooden

5.4.2

Tokyo Metropolitan Area 2705 2786

Miyagi Prefecture 2267 2090

Profitability of Earthquake-Proof Conversion of Rental Houses

Next, I provide the average willingness to pay an additional rent for the earthquakeproof conversion of rental houses such that the houses can withstand earthquakes with a seismic intensity of 7. Table 5.1 shows this willingness to pay—the average value of the willingness to pay for the earthquake-proof conversions of rental apartments (i.e., answer to questions in Fig. 5.3). Specifically, Table 5.1 shows the willingness to pay for the wooden and non-wooden apartments both in the Tokyo metropolitan area and Miyagi Prefecture. The figures in Table 5.1 times 12 correspond to X in Eqs. (5.1) and (5.2). The above finding shows that although 84% of the consumers express the need for the earthquake-proof conversion of rental houses, they are willing to pay only 2800 yen/month additionally for this conversion. In other words, there exists a gap between consumers awareness of earthquake-proof conversion of rental houses and their willingness to pay an additional rent for the conversion. As stated earlier, I estimate the NPV to determine the profits from this conversion. I estimate the numeric value of the willingness to pay X, from Table 5.1; I collect data on the lifespan of housing and estimate the discount rate and the cost of earthquake-proofing in Sect. 5.3.5. Based on the results of these estimations, I calculate the NPV of the earthquake-proof conversion of rental apartments, as shown in Table 5.2.

5.5 Factors Influencing the Willingness to Pay for Earthquake-Proof Conversion

91

Table 5.2 Net present value of earthquake-proof conversion of rental apartment (Yen/month) Wooden Non-wooden

Tokyo Metropolitan Area 1,053,839 1,381,308

Miyagi Prefecture 1,342,313 1,574,197

The earthquake-proof conversion of rental apartments has been less profitable in the Miyagi Prefecture than that in the Tokyo metropolitan area for both the wooden and non-wooden rental apartments. In other words, the NPV is negative, and the absolute value of the NPV is larger. This can be attributed to the fact that the willingness to pay for the earthquake-proof conversion of rental housing is smaller, and the cost of this conversion is higher. Higher cost reflects that the average area of rental apartments in the Miyagi Prefecture is larger than that of the Tokyo Metropolitan area. In both prefectures, the non-wooden rental apartments are less profitable than that of the wooden rental apartments; this can be attributed to the larger area and higher cost of non-wooden rental apartments. Overall, Table 5.2 indicates that the earthquake-proof conversion of rental housing is not profitable for rental firms. The outcome of NPV does not reveal an incentive for rental firms to make existing rental apartments earthquake-proof and enter the market of earthquake-proof rental housing.

5.5

Factors Influencing the Willingness to Pay for Earthquake-Proof Conversion

1. Household income Figure 5.5 shows that a higher household income induces a higher willingness to pay for the earthquake-proof conversion of both wooden and non-wooden rental houses. To analyze the relationship between the household income and the willingness to pay for the earthquake-proof conversion of rental housing, I verify the positive correlation between them. I perform a hypothesis test to determine whether the correlation coefficient between household income and the willingness to pay, that is, ρ1 is positive. The hypothesis of this hypothesis test is as follows: H 0: ρ 1 ¼ 0 H 1: ρ 1 > 0 I perform a two-sided test and present the outcome in Table 5.3. The hypothesis H0 is rejected at the 5% significance level, for both the wooden and non-wooden apartments. In other words, I find a positive correlation between the household income and the willingness to pay for the earthquake-proof conversion of rental houses.

5 Implementation of Subsidy for Improving the Earthquake-Proof Conversion. . .

92

Yen per month

6,000 5,000 4,000 3,000 2,000

wooden

1,000

non-wooden

0

Yen

Fig. 5.5 Household income and the willingness to pay for earthquake-proof conversion

Table 5.3 Correlation between household income and the willingness to pay

Wooden Non-wooden

t-statistics 3.163 3.663

2. Age of the respondents Figure 5.6 illustrates that the willingness to pay for the earthquake-proof conversion of the wooden and non-wooden rental houses is the lowest among the respondents in their fifties. Respondents in this age group face different financial pressures related to the educational expenses of their children, nursing care costs of elderly parents, and retirement preparedness. According to the survey of Japan’s local governments, an increasing number of aged households have been hampering the earthquake-proof conversion of rental houses (Asahi Newspaper, 2017). However, Fig. 5.6 shows that aged households are willing to pay more than those in their fifties for the earthquakeproof conversion of rental houses. Figure 5.5 shows that the willingness to pay is lower than the trend line for respondents with a household income of 8–10 million yen. The original data in this chapter shows that the proportion of respondents in the fifties age group among respondents with a household income of 8–10 million yen is 15.79% and is higher than the proportion of fifties in the whole respondents (6.45%). The low willingness to pay among the respondents in the fifties age group shows the low willingness to pay among respondents with a household income of 8–10 million yen. 3. Awareness of earthquake-proof conversion of housing The average willingness to pay differs between the respondents who replied “Yes” and those who replied “No” when they were asked about the need of

5.5 Factors Influencing the Willingness to Pay for Earthquake-Proof Conversion

93

5,000

Yen per month

4,500 4,000 3,500 3,000 2,500

wooden

2,000

non-wooden

1,500 1,000 500 0 twenes

thires

fores

fiies

sixes

sevenes

Fig. 5.6 Age of the respondents and the willingness to pay for earthquake-proof conversion

Table 5.4 The average willingness to pay and awareness of earthquake-proof conversion of housing (Yen/month)

Wooden Nonwooden

Respondents who are aware of earthquake-proof conversion of housing 2709 2661

Respondents who are not aware of earthquake-proof conversion of housing 1388 1405

earthquake conversion of rental housing to help withstand an earthquake with a seismic intensity of 7 (Table 5.4). The ratio of the average willingness to pay among respondents who replied “Yes” to that of the respondents who replied “No” is 1.95 and 1.89 for wooden and non-wooden rental apartments, respectively. This finding has implications for public relations (PR) activities that convey the significance and raise awareness of constructing rental apartments that withstand earthquakes with a seismic intensity of 7. The execution of these PR activities through various print and digital media (newspapers, magazines, community papers, internet articles) is expected to increase the consumers’ willingness to pay, consequently increasing the NPV of rental firms conducting such constructions. 4. Difference between the prefectures Despite being one of the disaster areas of the Great East Japan Earthquake, the willingness of Miyagi Prefecture’s respondents to pay is lower than that of the Tokyo metropolitan area’s respondents (Fig. 5.1). Given the positive correlation between the household income and the willingness to pay, this finding can be attributed to the fact that the average household income of respondents living in the Miyagi Prefecture (4.2683 million yen) is lower than that of respondents living in the Tokyo metropolitan area (4.9345 million yen).

5 Implementation of Subsidy for Improving the Earthquake-Proof Conversion. . .

94

5.6

Subsidy to Accelerate Earthquake-Proof Conversion of Rental Apartments

This section proposes a method to estimate the amount of subsidy necessary for the earthquake-proof conversion of rental apartments, based on the examples about the Miyagi Prefecture and the Tokyo metropolitan area. Firstly, I explain the current situation of subsidy for the earthquake-proof conversion of rental housing. Secondly, as a policy implication, I propose a plan for the formulation of local government subsidy for the earthquake-proof conversion of rental apartments. 1. Status of subsidy for earthquake-proof conversion of rental apartment Concerning the subsidy execution rate for the earthquake-proof conversion of rental apartment, it is 34% in the cities and wards of the Tokyo metropolitan area, and it is 0% in the local governments in the Miyagi Prefecture (as calculated by the author using the Japan Building Disaster Prevention Association (2017) and NTT Resonant (2017)). There is a significant variance in this subsidy amount among the local governments. For example, for the wooden rental apartments, the upper limit of this subsidy is 500,000 yen in the Komae city and 1,500,000 yen (the data is from the website of each local government). For the non-wooden apartments, this subsidy is 0 yen and 500,000 yen in the Higashi Murayama city and the Ota ward, respectively. Several local governments do not supply a subsidy for the earthquake-proof conversion of rental apartments. Moreover, this subsidy is not determined on the basis of the amount required by the rental firms executing the earthquake-proof conversion of rental apartments. 2. A plan for earthquake-proof conversion of rental apartment (a) Need for the subsidy Since the earthquake-proof conversion of rental houses has been unprofitable for rental firms, it is necessary to avail subsidy to accelerate this construction work. The governments should supply this subsidy for the following reasons. Firstly, it is crucial to protect inhabitants’ lives from earthquakes. Secondly, non-earthquake-proof rental apartments bring external diseconomies. In the case of non-earthquake-proof rental apartments, there has been a twofold increase in the percentage of fires caused by the collapse of rental apartments in large earthquakes and decline in inhabitants’ early fire extinguishing efforts (MLIT, 2000). Thirdly, the collapse of rental apartments causes road blockades, obstructing the passage of relief vehicles and fire engines. (b) Amount of subsidy I set the subsidy amount to motivate rental firms to carry out the earthquake-proof conversion of rental apartments. Subsidy helps earthquake-proofing projects for rental apartments earn positive NPV, and thereby motivates the rental firms to execute the projects promptly and enter the earthquake-proof rental housing market. In order to accelerate the earthquake-proof conversion of the rental apartments, the subsidy amount

5.6 Subsidy to Accelerate Earthquake-Proof Conversion of Rental Apartments

95

should fill a gap between the expected profit and cost. The amount of subsidy should be, at least, the absolute value of discounted present value of the present and future increment in rent minus the cost (presented in Table 5.2). Overall, the per unit subsidy amount should be 1,053,839 yen and 1,381,308 yen for wooden and non-wooden apartments, respectively, in the Tokyo metropolitan area; it should be 1,342,313 yen and 1,574,197 yen for the wooden and non-wooden apartments, respectively, in the Miyagi Prefecture (c) Method to determine the need for and amount of subsidy Any local government can adopt the above method to estimate the subsidy amount required to accelerate the earthquake-proof conversion of rental apartments in the Miyagi Prefecture and the Tokyo metropolitan area. This method assesses if and how much subsidy is required by the rental firms. If a rental firm qualifies to receive a subsidy, the subsidy’s amount is calculated based on the firm’s need. This method involves the following steps. First, using the CVM, I calculate the consumers’ willingness to pay for the earthquake-proof conversion of rental apartments. Second, using the outcome of the willingness to pay, I assess the need for subsidy according to the rental firms’ NPV, which is calculated using the DCF method. A negative NPV shows the unprofitability of the earthquake-proofing project and highlights the need for a subsidy. I calculate the minimum amount of the subsidy as the absolute value of NPV. (d) Institutional Design of Subsidy Since respondents earning a low household income and those in their fifties have a low willingness to pay for the earthquake-proof conversion of the rental apartments, these respondents should be allocated a higher subsidy. For instance, the subsidy amount can be set by assuming the average inhabitants, as shown in (c), and subsequently adding the subsidy to apartments rented out to the low-income households and those in the fifties. (e) Proposal for Subsidy Concerning the above method, first, I propose a subsidy of local government. This is because inhabitants’ lives form the top priority of a local government’s subsidy (Hirono, 2020a). In this context, it must be noted that the Ministry of Internal Affairs and Communications (2017, 2018, 2019) has indicated an increase in the amount of prefectural tax in Japan since 2011. This trend has also been observed in the disaster areas of the Great East Japan Earthquake—the Miyagi, Fukushima, Ibaraki, and Tochigi Prefectures. Second, about local governments that cannot provide a subsidy for earthquake-proofing, I suggest a tax. For example, a reduction of property tax on rental apartments can contribute toward the execution of the earthquake-proof conversion project; the amount of assistance through this reduction in the property tax can be calculated using the method in (c). The calculation of property tax for each rental apartment varies by each situation, and this tax can be levied on each owner of the rental apartments.

96

5.7

5 Implementation of Subsidy for Improving the Earthquake-Proof Conversion. . .

Conclusion

Although over 80% of the respondents expressed a need for the earthquake-proof conversion of apartments, they were willing to pay only less than 3000 yen/month in addition to the present rent for the earthquake-proof conversion. For the rental firms, this implies negative profits from such construction projects. This finding shows the lack of progress in the earthquake-proof conversion of rental apartments. Given this situation, there is a need for subsidies to make houses more earthquake-proof in order to avoid the external diseconomies and loss of human lives. To address this issue, I constructed a framework to calculate the amount of municipal governments’ subsidy. To this end, I calculated the difference between the discounted present value of the present and future cash flows minus the cost of the earthquake-proof conversion of rental housing, using the CVM and DCF methods. This framework can be applied to any municipal government subsidy program for the earthquake-proofing projects. The findings also show that the willingness to pay for the earthquake-proof conversion of rental apartments depends on the household income, respondents’ age, and the awareness of the need for earthquake-proof conversion of houses.

References Alonzo, F. (2002) “The Benefits of Building Barrier-free: A Contingent Valuation of Accessibility as an Attribute of Housing,” European Journal of Housing Policy, 2(1), 25–44. Asahi Newspaper (2017) “Earthquake-Proof Conversion of Housing; Unattainable Goal,” the January 16 issue. Cameron, T. A. (1988) “A New Paradigm for Valuing Non-market Goods Using Referendum Data: Maximum Likelihood Estimation by Censored Logistic Regression,” Journla of Environmental Economics and Management, 15(3), 355–79. Ciriacy-Wantrup, S. V. (1947) “Capital Returns from Soil-conservation Practices,” Journal of Farm Economics, 29, 1181–1196. Cohen, M. A., R. T. Rust et al (2004) “Willingness-to-pay for Crime Control Programs,” Criminology, 42(1), 89–110. Davis, R. K. (1963) “The Value of Outdoor Recreation: an Economic Study of Maine Woods,” Unpublished Ph.D. Dissertation. Harvard University, Cambridge, MA Duffield, J. W. and D. A. Patterson (1991) “Inference and Optimal Design for a Welfare Measure in Dichotomous Choice Contingent Valuation,” Land Economics, 67(2), 225–39. Hanemann, W. M. (1984) “Welfare Evaluations in Contingent Valuation Experiments with Discrete Responses,” American Journal of Agricultural Economics, 66, 332–41. Hidano, N. (eds) (1999) Kankyo to Gyosei no Keizaihyouka—CVM manual—[Economic Evaluation of Environment and Administration: Manual of CVM]. Tokyo: Keiso Shobo. Hirono, K. N. (2004) “Imperfect Information and the Amount of Housing Ownership,” Pacific Economic Review, 9(4), 335–43. Hirono, K. N. (2019) “Kasoshijouhou niyoru Jutaku no Hyouka (Evaluation of Housing Facility by the Contingent Valuation Method),” Journal of Property Assessment Polily, 19(2), 1–8.

References

97

Hirono K. N. (2020a) “Jutaku no Taishinka ni Muketa Hojokin—Taishinka heno Shohisha no Shiharaiishigaku no Bunseki ,” in Hirono K. N. and K. Yaguchi (eds) Higashinihondaishinsai kara 10 Nen: Saisei Hatten ni Okeru Kadai no Bunseki (Analysis on Reconstruction Period and the Stages of Regeneration after a Great Earthquake: Ten Years after Great East Japan Earthquake), Tokyo: Taisei Shuppan. Hirono, K. N. (2020b) “Chintaijutaku no Taishinka ni Muketa Hojogaku no Suikei (A Method for Estimation of the Amount of Subsidy to Accelerate Earthquake-Proof Conversion of Rental Housing),” Studies in Regional Science, 50(1), 19–38. Japan Building Disaster Prevention Association (2017) Chihokokyodantai niokeru Taisinsindan Kaishu no Shienseido (Local Governments’ Support System for Seismic Diagnosis and Earthquake-Proof Conversion), http://www.kenchikubosai.or.jp/soudan/sien.html Japan Meteorological Agency (2021) Mechanism of Earthquake Occurrence, http://www.data.jma. go.jp Kuriyama, K. (1998) Kankyo no Kachi to Hyoka no Shuho—CVM niyoru Keizaihyouka—(Value of Environment and a Method of Estimation), Sapporo: Hokkaido University. Ministry of Internal Affairs and Communications (2017, 2018, 2019) White paper on local finance, Tokyo. Ministry of Internal Affairs and Communications, Various issues, Prefectural Financial Results, Tokyo. Ministry of Land, Infrastructure, Transport and Tourism (1996) Construction White Paper, Tokyo. Ministry of Land, Infrastructure, Transport and Tourism (2000) Taishinkaishusuishin Chosa no Kekkka (Outcome of Survey on Acceleration of Earthquake-Proof Conversion), Tokyo. Ministry of Land, Infrastructure, Transport and Tourism (2017) Jutaku Shijo Doko Chosa Houkokusho (A report on the survey of the Japanese housing market), Tokyo. Morita, T. and S. Managi (2018) “Model Comparisons on Estimating Willingness to Pay for Auto Driving Systems,” in: Yamanashi Glocal Studies, Yamanashi Prefectural University, 13, 71–80. National Police Agency (2020, 2011) Police Activities and Damage Activities in the Great East Japan Earthquake in 2011, Tokyo: Emergency Disaster Security Headquarters in the National Police Agency. New Japan Reform (2019) Earthquake-resistant Construction of Building and Apartments, http:// www.snreform.co.jp/reform/earthquakeproof/ NTT Resonant (2017) Data on living in the Tokyo Metropolitan Area, https://house.goo.ne.jp/ chiiki/kurashi/ Ruszkowski, R. (2017) “The Contingent Valuation Method in Assessing the Value of Sport’s Stadium in Developing Nations. The Case of Poland.,” MPRA Paper, No.80581, 1–20. Sato, K. (2018) “Estimation of Housing Earthquake Resistant Rate Using Micro Data of Housing and Land Survey 2013,” Research Paper, No.42, Statistical Research and Training Institute, Ministry of Internal Affairs and Communications. Sato, K. and M. Tamamura (2006) “A Study on Resident Consciousness for Retrofitting of Existing houses by Contingent Valuation Method: A Case Study at Ichikawa City, Chiba Prefecture,” Journal of Social Safety Science, No.8, Institute of Social Safety Science, 81–87. Sato, S., M. Kondo and K. Watanabe (2005) “Analysis of Demand for Barrier-free House and Effect in Reduction of Burden,” Journal of Architecture and Planning, No.592, 193–99. Statistics Bureau of Ministry of Internal Affairs and Communications (2017) Results of 2013 Housing and Land Survey, Tokyo. Taishin Net (2017) Initial Cost of Seismic Reinforcement, http://www.taisin-neto.com/solution/ online_seminer/sindanhokyou/ Taniguchi, H. (2016) “Shinsai to Community (Earthquake and communities),” in Maruo, N., K. Yaguchi and G. Miyagaki (eds) Community no Saisei (Regeneration of Communities), Tokyo: Chuokeizai-sha.

98

5 Implementation of Subsidy for Improving the Earthquake-Proof Conversion. . .

Taylor, C. B., M. Stevenson, S. Jan et al. (2010) “A Systematic Review of the Costs and Benefits of Helicopter Emergency Medical Services,” Injury, 41, 10–20. Yamaga, H., M. Nakagawa and M. Saito (2002) “Earthquake Risk and Rent: Political Implication for Earthquake Proof Countermeasure,” JCER Economic Journal, 46, 1–21. Yokochi M., T. Terashita and K. Ogasawara (2014) “Comparison of the Willingness to Pay for Air Ambulance between Three Regions Using CVM,” Journal of the Japan Society for Healthcare Administration, 51(1), 41–52. Yoshizawa Architectural Structure Design (2013) Tateru [Construct]. Koshigaya.

Chapter 6

Housing Policy to Supply Barrier-Free Rental Housing

Abstract In Japan, there is a desperate need for barrier-free rental housing because of the rapidly aging population, and because barrier-free rental housing presently only accounts for 32.68% of the total rental housing stock. In this chapter, I propose a policy to solve the problems associated with the supply of barrier-free rental housing. Our scheme involves a method of funding rental housing with a securities investor, as well as implementing property management and introducing rent subsidies. The plan entails funds being collected to construct barrier-free rental housing. This project will generate positive rental profits, which would, in turn, lead to an increase in the supply of barrier-free rental housing. According to the plan, the low-income elderly will also have access to barrier-free rental housing. Keywords Barrier-free housing · Securities investor · Aging of the population · Rent subsidies · The low-income elderly

6.1

Introduction

Japan’s housing loan policy facilitated an expansion of the housing stock, in the sense that lowering the interest rate on Government Housing Loan Corporation loans increased housing starts as shown in the previous chapter. The quantity of Japanese housing is satisfied. Nowadays, the ratio of vacant houses to total houses has increased to 13.6% in Japan (Statistics Bureau, 2018). The purpose of the housing policy is now to increase the quality of housing, rather than the quantity. Japan has a rapidly aging population. I define the elderly as people aged 65 or over. The ratio of the elderly to the total population (over-65 ratio) is now 27.7% in Japan (National Institute of Population and Social Security Research, 2019). However, the stock of barrier-free housing—especially housing that is appropriate for the elderly—is insufficient, as noted in Sect. 6.3. In this chapter, I firstly consider aging of the population in Japan. Secondly, I demonstrate the need to resolve the shortage of housing that is suitable for the elderly. Finally, I present a housing policy plan to solve some of the disincentives relating to the supply of appropriate housing for the elderly. © Springer Nature Singapore Pte Ltd. 2022, corrected publication 2022 K. N. Hirono, Economic Analysis of Housing Policy in Japan, New Frontiers in Regional Science: Asian Perspectives 64, https://doi.org/10.1007/978-981-19-4925-8_6

99

100

6

Housing Policy to Supply Barrier-Free Rental Housing

The housing policy presented in this chapter involves supporting barrier-free rental housing by (1) proposing a method of funding in which a government sector becomes a securities investor, and (2) suggesting property management of housing for the elderly and that rent subsidies for the low-income elderly tenants be offered. I demonstrated that this policy would produce several positive effects including an increase in construction of barrier-free rental housing, through the development of a model. This is the first research to propose such a policy and to construct a model of barrier-free housing. The previous studies that have analyzed the housing market with respect to Japan’s elderly population are as follows. Hayakawa (1990) points out that highquality housing stock is necessary for the elderly to maintain their independence, and live without excessive nursing care. From Japanese case studies, Hayakawa (1990) shows that the poor structure and the narrowness of housing in Japan cause problems such as illness, inability of the elderly to live independently, and difficulties in receiving nursing care at home from families and home care workers. Kajima (2002) presents the results of a survey to show the present situation of housing reform using Japanese nursing care insurance system. The survey includes costs and descriptions of housing reforms, parts in houses where reforms were implemented, and consumers’ insufficient information about housing reform. Ministry of Construction Policy Research Center (1993) calculates the reduction in the costs of future nursing care as a result of the construction of barrier-free housing and concludes that making houses barrier-free would significantly reduce costs of nursing care and medical care of the elderly. They also find that the effectiveness of making a Type 1 house barrier-free is 5.2 times the cost and that of making a Type 2 house barrier-free is only 1.1 times the cost, where Type I housing is accessible for those using a walking stick and is equipped with fundamental barrier-free equipment, such as handrails, a lack of steps, and bathtubs that are easy to step into, and Type II housing are designed to suit elderly in wheelchairs, with slopes rather than stairs and a wheelchair lift at the entrance in addition. The result of this research supports the fundamental points of views of my research by showing the needs to construct barrier-free housing. However, these studies do not recommend concrete policies that could overcome the negative effects of the present situation. Moreover, they do not use models in their analysis. These factors therefore form the primary points of differences between previous researches and the analysis undertaken in this chapter. This chapter proceeds as follows. In Sect. 6.2, I consider the aging of the population in Japan. In Sect. 6.3, I demonstrate the need to resolve the shortage of housing that is suitable for the elderly. In Sect. 6.4, I summarize the present situation of barrier-free housing in Japan. In Sect. 6.5, I describe the conditions that hamper the supply of barrier-free rental housing. In Sect. 6.6, I present a housing policy that could remove many of the disincentives relating to the supply of barrier-free rental housing. In Sect. 6.7, I demonstrate the effects of the proposed policy. Section 6.8 concludes.

6.2 Overview of Population Aging in Japan

(%)

Actual Values

101

Estimated Values

Japan Sweden Germany France United Kingdom United States

Year Fig. 6.1 Transition of the ratio of the elderly (over-65 in developed countries). Note: the percentages in year 2015 in brackets. (Source: Cabinet Office (2019))

6.2

Overview of Population Aging in Japan

As shown by Fig. 6.1, Japan’s over-65 ratio reached 26.6% in 2015, which is the highest ratio of any developed country. Japan’s over-65 ratio has also been growing rapidly. It stood at 4.91% in 1950, increased to 19.9% in 2005, and is expected to rise to 36.37% by 2050 (National Institute of Population and Social Security Research, 2019). As for Japan’s over-65 ratio, not only the ratio is high but also the doubling time—the doubling number of years of over-65 ratio from 7% to 14%—is only 24 years, indicating the exceptional pace of this population aging. For comparison, Germany’s doubling time is 40 years, the United Kingdom’s is 46 years, the United states’ is 72 years, Sweden’s is 85 years, and France’s is 115 years. Japan’s high

102

6

Housing Policy to Supply Barrier-Free Rental Housing

speed of population aging is reflected by the steepness of the over-65 ratio’s curve in Fig. 6.1. The over-65 ratio is rising so quickly because the elderly population is growing while the total population is decreasing because of a falling birthrate. The elderly population was 22,005,000 in 2000, and it increased rapidly to 35,152,000 in 2017. Simultaneously, Japan’s total population fell from its peak of 128,057,000 in 2010 to 126,706,000 in 2017. The total population will continue to decrease from now on and is estimated to fall to be 101,923,000 in 2050, according to the National Institute of Population and Social Security Research (2019). The rise in the number of elderly people in Japan is due to a decrease in mortality rates and a rise in average lifespans, which are largely driven by advances in medical technology and Japanese diets (Statistics Bureau, 2018). Data from the Ministry of Health, Labor and Welfare (2018a) shows that the average lifespan is 81.25 for a Japanese man and 87.26 for a Japanese woman. As a result, the average lifespan of both Japanese man and woman has been extended by 30 years in only a time period of 70 years. Moreover, this level of average life span (that is average of Japanese male and female is 84.26) is top in the world. Moreover, aggregated (i.e., average of Japanese male and female) lifespan is 84.26 and is highest in the world. Presently, cancer is responsible for the largest proportion of deaths in Japan (Ministry of Health, Labor and Welfare, 2018b). However, recent advancement in medical technology which almost cures cancer has further raised life expectancy. Japan’s medical system is one of the most extensive in the world. Japan has a national health insurance program ensuring that medical care is readily available. In addition, there is a system of medical examination of the elderly, which enables the early detection of problems and immediate treatment in hospitals (The Japan Foundation for Aging and Health, 2020). The Japanese diet called Washoku has played a large part in placing Japan at the top of global tables for average lifespan (Ministry of Agriculture, Forestry and Fisheries, 2018). Japanese food is low in calories, consisting mainly of rice, plant protein, seafood, vegetables, fruit, seaweed, and milk in well balance. Japan’s total fertility rate was only 1.42 in 2018—well below the natural replacement rate. The explanatory factors for this include decrease in marriage rates, a tendency to marry later, and women’s increasing participation in the workforce. Environment for workers to strike a balance between childcare with work is not developed in Japan, which is a large cause that decreases birthrate (Statistics Bureau, 2018). Furthermore, Suzuki et al. (2007) find a statistically significant negative correlation between educational expenditure and birthrate, implying that a focus on educational attainment is one factor that lowers the birthrate.

6.3 The Need to Supply Housing Suitable for the Elderly

6.3

103

The Need to Supply Housing Suitable for the Elderly

As shown in Sect. 6.2, Japan now has the world’s highest percentage of elderly in its population and this ratio continues to increase at an alarming rate. Therefore, the issue of insufficient barrier-free housing must be urgently addressed. It is necessary to supply sufficient housing suitable for the elderly for the following reasons. Firstly, the elderly reside in 41.99% of the total housing in Japan (calculated by the author, using the data provided by the Statistics Bureau (2018)). Secondly, barrier-free housing reduces the cost of nursing care for both the country and individual households. For example, with barrier-free housing, the elderly are able to move by themselves in corridors and steps without home care workers’ aid, and the elderly can take baths without home-visit bathing services (Ministry of Construction Policy Research Center, 1993). Thirdly, increasing barrier-free housing will attain national minimum by making houses safe for the elderly. Fourthly, making housing barrier-free means homes become more appropriate places in which to deliver nursing care. Fifthly, cohabitation rate for children with elderly people was about 70% in 1980, but it largely decreased to 39% in 2015. In 1980, a little less than 30% of elderly people lived in “one-person households” or “households of only a couple,” but this percentage rose to 56.9% in 2015 (Statistics Bureau, 2018). It is natural to think that, as a result, when a trouble caused by some barrier in the house occurs to the body of the elderly person (for example, an elderly person almost tripping over or getting into difficulty in a bathtub because it is not accessible), the elderly person’s son or daughter cannot soon come to rescue him (or her) on average. Considering this, houses should be made barrier-free so that the elderly can live in houses independently without facing the risks caused by barriers. Finally, while younger people can live in houses that lack accessibility with their ability to adapt to circumstances, the elderly cannot and this causes accidents in houses. Thus, the elderly require barrier-free housing (Ito and Sonoda, 1994). According to the Ministry of Health, Labor and Welfare (2010), 85.38% of all deaths caused by domestic accidents in Japan can be attributed to the elderly. The main cause of accidental deaths of the elderly at home is drowning in the bath (30.97%), followed by slipping and falling on the floor (11.18%) and falling from a step or stairs (2.77%) (Ministry of Health, Labor and Welfare, 2010). Therefore, bathtubs that are easy to step into, no steps indoors, and handrails being placed along stairs can prevent many accidental deaths of them and make houses safer for the elderly. Additionally, nursing care is easier to provide with “hallways wide enough for a wheelchair” and “indoors without steps.” Therefore, making housing barrierfree, with bathtubs that are easy to step into, no steps indoors, and handrails along stairs and hallways wide enough for a wheelchair, is necessary.

104

6

Housing Policy to Supply Barrier-Free Rental Housing

6.4

The Current State of Barrier-Free Housing

According to Table 6.1, the ratio of barrier-free housing to the total housing stock in Japan is 50.86%. This ratio is only 32.68% for rental housing, but it is 64.14% for owner-occupied housing. The individual ratios by item in rental housing are low: Only 8.55% of houses have the “bathtubs easy to step into,” 8.97% have “hallways wide enough for a wheelchair,” 13.86% have “indoors without steps” and 9.82% have “handrails along stairs.” Furthermore, the ratio of rental houses that satisfy all four of these conditions to the total rental houses is only 8.55%. Therefore, in this chapter, I present a plan to increase the number of barrier-free rental houses.

6.5

Factors That Discourage the Supply of Barrier-Free Rental Housing

There are three main reasons why the supply of barrier-free rental housing is being restrained. The first cause is imperfect information, the second is that the evaluation of a collateral value of rental houses does not increase when they are made barrierfree, and the third is that there are many elderly people who cannot afford the rent of barrier-free housing. The first and second causes relate to the market mechanism, whereas the third cause must be solved through a welfare policy. As for the first cause, there is little information regarding whether rental housing is barrier-free or not in advertisements placed by real estate agents and in housing informational magazines, such as Shukan Jutaku Joho (renamed to SUUMO). Therefore, rents for barrier-free rental housing are the same as rents for non-barrierfree rental housing, because consumers evaluate barrier-free rental housing in the same manner as non-barrier-free rental housing.1 However, additional costs of 5.3% on average (Foundation for Senior Citizen’s Housing (1998)) must be borne by the suppliers for constructing barrier-free housing, as compared to making new-built houses with barriers. The ratio of amount which is passed on to the consumer to the cost of making housing barrier-free is under 100%.2 Therefore, making rental houses barrier-free reduces the profits of suppliers, thus barrier-free rental housing is rarely supplied. Setting the rent of barrier-free rental housing at a level that reflects the cost of making rental housing barrier-free, supplying barrier-free rental housing would not eliminate the profits of suppliers. However, interviews conducted with officials from the Japan Housing Finance Agency and city bankers in Tokyo suggest that, even if this were to be the case, banks would still be wary of providing loans to suppliers because banks are uncertain over occupancy rates given the prevalence of imperfect

1 2

This information is taken from Foundation for Senior Citizen’s Housing. This information is taken from Foundation for Senior Citizen’s Housing also.

6.61

10.86 10.99

2.12

2.16

9.82 0.90

Dwelling Corridor Stairs rooms 5.72 26.22 1.40 8.10 37.14 1.77

Source: Calculated by the author using data provided by Statistics Bureau (2018)

Total Owned housing Rental 100.00 32.68 21.67 housing

Dressing Total Total Total Entrance Toilet Bathroom room 100.00 50.86 41.75 12.06 20.74 23.32 3.22 100.00 64.14 55.65 15.87 27.59 31.73 4.04

Equipped with handrail

Facilities for the elderly

Table 6.1 Proportion of housing that is barrier-free for the elderly (%)

0.52

8.55

Others 0.92 18.78 1.21 25.73

8.97

15.52 20.15

Bathtubs Hallways wide easy to step enough for into a wheelchair

13.86

20.94 26.17

Indoors without steps

6.5 Factors That Discourage the Supply of Barrier-Free Rental Housing 105

106

6

Housing Policy to Supply Barrier-Free Rental Housing

information. Occupancy rates could decrease with the higher rents, since it is difficult for consumers to obtain information which convey that this rental housing is barrier-free. Hesitation of banks to issue loans makes it difficult for suppliers to accumulate sufficient construction funds. If there are no new financiers to accept the risk of undertaking barrier-free housing projects, and there is no measure to overcome imperfect information, banks will refrain from issuing loans for barrier-free rental housing. The second cause is that appraisal values of barrier-free rental housing will not rise,3 since rents of them will not rise by being barrier-free. Thus, banks will not issue the additional loans that are necessary for suppliers to pay the costs associated with making rental houses barrier-free, which acts as a disincentive for suppliers to construct barrier-free rental housing. Thus, there is a need for new financiers to cover the costs to construct barrier-free rental housing.4 Third, approximately 89% of the elderly who live in rental housing perceive housing costs as a significant financial burden (Ministry of Land, Infrastructure, Transport and Tourism, 2015). For these households, increasing rental costs for barrier-free housing will exert an increased pressure on their budget. Therefore, many of these households are unable to move into barrier-free rental housing.

6.6

Housing Policy Plan to Increase Barrier-Free Rental Housing

I propose a plan to enable the supply of more barrier-free rental housing. The plan consists of rent subsidy for the low-income elderly and securities investment to support construction fund of barrier-free rental housing which the high-income elderly live in. In addition, property management is an important component of this plan, to resolve incomplete information. I set four criteria for rental housing, namely, “bathtubs easy to step into,” “indoors without steps,” “handrails along stairs,” and “hallways wide enough for a wheelchair.”

3

This information is stated in interviews with the Japan Housing Finance Agency and city bankers in Tokyo. 4 We can think of a case in which a supplier pays the cost to make rental housing barrier-free. However, this is not feasible in the present real estate market, according to interviews conducted with the Japan Housing Finance Agency and city bankers in Tokyo. If many owners can do this, there exist more barrier-free rental houses, which is not happening. Analysis in Sects. 6.3–6.7 is from Hirono (2009). I have added analysis of aging of the population in Japan in Sect. 6.2.

6.6 Housing Policy Plan to Increase Barrier-Free Rental Housing Assets

107

Liabilities

beneficial interest in trust Banks fund

Owner firm of a rental house (Originator)

Land money securities A Government sector

beneficial assets

fund interest in trust

Trust banks

SPC Fig. 6.2 A method of funding of barrier-free rental housing

6.6.1

Method of Funding and Property Management

I propose a housing policy plan which involves one government sector becoming a securities investor for barrier-free rental housing to finance the costs to construct barrier-free rental housing. That is, as a securities investor, the government sector invests as much as the cost of making rental housing barrier-free and supports property management. The rental housing owner firm becomes the originator and places its assets in a trust bank. The trust bank issues trust beneficiary right to the owner firm. The originator transfers trust beneficiary right to the special purpose company (SPC), and the price of trust beneficiary right is the present discounted value of cash flow of this asset in the future. The SPC issues securities to the government sector, borrows money from a bank, and pays these funds and money to the originator. The securities issuance is backed by trust beneficiary right. A bank loan is included because of its leverage effect (Fig. 6.2). The order of priority to pay dividends from the profits from the rental housing’s rents and to distribute residual property after bankruptcy is: (1) bank loan, (2) securities investment of the government sector, and (3) most subordinated fund, which is the last fund to receive dividend (in a usual case) and the remaining property (in a case of bankruptcy). The owner firm owns the most subordinated fund. Using a securities investor and a most subordinated investor in the scheme provides a credit enhancing measure for bank loans. The average profit of a securities investor is positive, as will be shown below. Property management involves the promotion and maintenance of rental housing. Property management provides information to elderly potential tenants regarding whether the rental house is barrier-free or not and regarding the value of barrier-free

108

6

Housing Policy to Supply Barrier-Free Rental Housing

rental housing. Promotion is paid for with higher rental income, which also covers the costs of making rental housing barrier-free. Property management will enhance future rental profits, compared to general rental housing (without barrier-free housing), because the elderly who demand barrier-free rental housing evaluate barrierfree rental housing higher than general rental housing. I propose that the government engage in securities investment and supporting owner firms’ property management so that the rental housing program is successful. The purpose of property management is to resolve imperfect information so that the rental houses collect enough tenants with higher rents, which cover the additional costs of barrier-free housing, and which will be paid to the government sector (a securities investor).

6.6.2

Rent Subsidy

Of the household which includes the elderly who live in rental housing, 75.31% earn below 3 million yen annually (calculated by the author, using the data from Statistics Bureau (2018)). For these low-income household including the elderly, I propose that rent subsidies be provided to compensate for the additional costs associated with making rental housing barrier-free. A government sector in charge of social security would execute this policy. The reason is that the low-income elderly could not afford barrier-free rental housing even if property management and the above method of funding were implemented.

6.7

Effects of the Policy

The effects of the above policy are as follows in detail: 1. A construction fund for barrier-free rental housing can be collected. This is because funds needed for the costs of barrier-free rental housing will be supplied by the government, as a securities investor, and because a bank will make loans, since a securities investor and an owner of most subordinated fund (the owner firm) will take on the risk of the barrier-free rental housing project. Even in the case of bankruptcy, there is a high possibility that the original principal of bank loans will be paid back because of the existence of a securities investor and an owner of most subordinated fund in the scheme. The bank will make loans for barrier-free rental housing, and owners of rental houses will be able to supply barrier-free rental housing because they can collect construction funds. Therefore, the supply of barrier-free rental housing will increase. 2. The low-income elderly also can live in barrier-free rental housing thanks to rent subsidies. This will increase the demand for barrier-free rental housing. 3. The sign of profits of barrier-free rental housing will change from negative to positive; thus, the owners of rental housing will construct more barrier-free rental

6.7 Effects of the Policy

109

housing. Rents will go up with property management. This is because, for the elderly who demand barrier-free rental housing, evaluation of barrier-free rental housing is higher compared to general rental houses, and the elderly will pay higher rents when the rental houses are promoted as being barrier-free. Promotion also has the effect of conveying the value of barrier-free rental housing. To observe the profits and attitudes of the owners of rental housing, I compare the cases of an owner who builds a general rental house to an owner who builds a barrier-free rental house. Construction cost including land is P0 for a general rental house and P0 + C0 for a barrier-free rental house per unit (i.e., cost of barrier-free is C0). The owner of a barrier-free rental house raises rent by x% (x > 0) by making a rental house barrier-free. Rent at time t is Rt (t ¼ 1, 2, . . .) for a general rental house, and rent is Rt (1 + x/100) for a barrier-free rental house. Durable years for both a general rental house and a barrier-free rental house are T. Let the remaining value at year T be STc for a general rental house and STb for a barrier-free rental house. The cost to scrap after use is gTc for a general rental house and gTb for a barrier-free rental house at time T. I assume gTb ¼ gTc because cost to scrap a house will not vary much whether steps and handrails exist or not. Let r denote discount rate. The owner plans to build condominiums as a whole. However, I will show calculations for one unit of rental house for simplicity. I assume monopolistic competition in the rental market. Price of a general rental house is the same as the average cost (cost per unit) at the long run equilibrium, and the price is P0. I assume the price is equal to the bank appraised price for a general rental house. A bank issues loan as much as 0.7 P0 (as common practice in the market), which is 70% of the value of a rental house. The owner of a rental house pays the rest of the cost, which is 0.3 P0. I suppose an owner thinking about construction of new barrier-free rental house. The rate of return on rental house is shown on a yearly basis. For simplicity, I omit analysis of tax. I evaluate the rental house using the present discounted value of its future profit. Therefore, I have this evaluation, P0 ¼

T X

ð1=ð1 þ rÞÞt Rt :

ð6:1Þ

t¼1

The profit from a barrier-free rental house is: πb ¼

T X t¼1

ð1=ð1 þ rÞÞt Rt ð1 þ x=100Þ  ðP0 þ C0 Þ þ ðSTb  gTb Þ=ð1 þ rÞT : ð6:2Þ

The profit from a general rental house is

110

6

πc ¼

T X t¼1

Housing Policy to Supply Barrier-Free Rental Housing

ð1=ð1 þ rÞÞt Rt  P0 þ ðSTc  gTc Þ=ð1 þ rÞT :

ð6:3Þ

As the remaining value of the rental house is the value of land after T, I have STb ¼ STc . Let A ¼ π b  π c. I obtain Eq. (6.4) because gTb ¼ gTc . A¼

T X

ð1=ð1 þ rÞÞt Rt ðx=100Þ  C 0

ð6:4Þ

t¼1

¼ P0 ðx=100Þ  C 0 ¼ P0 ðx=100  C 0 =P0 Þ < 0:

ð6:40 Þ

I have an inequality in Eq. (6.4) because the rate of increase in cost as a result of making a rental house barrier-free exceeds the rate of increase in rent. The cause of this is imperfect information about whether rental housing is barrier-free or not as shown in Sect. 6.5. Since making a rental house barrier-free decreases profit, an owner chooses not to make a rental house barrier-free. I can make the level of x/100 in Eq. (6.4) larger than C0/P0 through property management by conveying that the rental house is barrier-free and importance of barrier-free housing to consumers. Therefore, the owner’s profit rate for making a rental house barrier-free will be positive, which is negative without property management. The property management will encourage the owners of rental houses to decide to construct barrier-free rental house because of this positive profit rate. Thus, the number of barrier-free rental house will increase. 4. I will now show how the average dividend and the average rate of return of the securities investor will go up with property management. I can presume that a bank provides α1 of a barrier-free rental house’s construction funds, a government sector provides α2 as a securities investor, and the owner of the rental house supplies α3, as an owner of most subordinated fund (α1 + α2 + α3 ¼ 1, 0 < α1, α2, α3 < 1). Therefore, the amount of the bank loan is α1(P0 + C0), that of the securities investment is α2(P0 + C0), and that of the owner’s fund is α3(P0 + C0). For simplicity, let the interest rate of the bank loan be 0% (the substance of the argument will not change with this assumption). Because of the leverage effect realized by low-interest rate of bank loan (here it is 0%), profit (which will be delivered to a securities investor and the owner of the firm) of a barrier-free rental house each year will go up. In addition, property management works. If α1 ¼ 1/2, the dividend of a barrier-free rental house for a securities investor and the owner of the rental house will be 2Rt (1+x/100) per unit. As is common practice in the market, a bank lends 0.7 P0 (that is, 70% of evaluation of a barrier-free rental house) and the owner invests the remaining 0.3 P0. Let the cost of making a house barrier-free be 10% of the total construction cost, and the securities investor (the government sector) pays this cost. The dividend to a securities investor and to the owner of the firm is 11/4Rt (1+x/ 100) in total. The leverage coefficient is 11/4. In general, we can set C0 ¼ θP0.

6.7 Effects of the Policy

111

The total amount of the dividend to a securities investor and the owner of the firm is (10 + θ)/(3 + θ)Rt (1+x/100) and the leverage coefficient is (10 + θ)/(3 + θ). As for a barrier-free rental house, I suppose that the occupancy rate is y% (0  y  100). I set leverage coefficient as 11/4, as is common value in the market, then rental income per unit is (y/100)(11/4)Rt (1+x/100) on average. I assume that when the occupancy rate is higher than (or the same as) y*, profits will be delivered to both a securities investor and the owner of the firm. If the occupancy rate is lower than y*, profits will be delivered to a securities investor only. z% denotes the maximum rate of return of a securities investor. The owner of the firm can receive profits in the case that the occupancy rate is higher than (or equal to) y*. The occupancy rate of a barrier-free rental house is y1 in a good case (high rate of occupancy of a barrier-free rental house), and y2 in a bad case (low rate of occupancy of a barrier-free rental house) (y2 < y1). Probability of a good case is τ1 and of a bad case is τ2 (0 < τ1, τ2  1, τ1 + τ2 ¼ 1). There are three cases for the occupancy rate: (a) y*  y2 < y1  100 (b) 0  y2 < y* < y1  100 (c) 0  y2 < y1 < y* For each case, I look at the dividend and the average rate of return of a securities investor. We let θ ¼ 0.7. The following argument does not change even if the leverage coefficient is (10 + θ)/(3 + θ), namely, the general case. (a) Suppose first that y*  y2 < y1  100: The rate of return of a securities investor is z% for both a good case and a bad case. Therefore, the average rate of return of a securities investor is z%, and his or her dividend is z[α2(P0 + C0)] for both cases and on average. The rate of return of the owner of the firm for a good case is {[(y1/100) (11/4)Rt (1+x/100)]  z[α2(P0 + C0)]}/[α3(P0 + C0)] and for a bad case is {[(y2/100)(11/4)Rt (1+x/100)]  z[α2(P0 + C0)]}/[α3(P0 + C0)]. His average rate of return is τ1{[(y1/100)(11/4)Rt (1+x/100)]  z[α2(P0 + C0)]}/ [α3(P0 + C0)] + τ2{[(y2/100)(11/4)Rt (1+x/100)]  z[α2(P0 + C0)]}/ [α3(P0 + C0)] (b) Suppose that 0  y2 < y* < y1  100: The rate of return of a securities investor is z% for a good case and [(y2/ 100)(11/4)Rt (1+x/100)]/[α2(P0 + C0)] for a bad case. That is, a securities investor will receive divisions of profits by priority for a bad case. The average rate of return of a securities investor is τ1z + τ2[(y2/100)(11/4) Rt (1+x/100)]/[α2(P0 + C0)]. The dividend of a securities investor is z [α2(P0 + C0)] in a good case and [(y2/100)(11/4)Rt (1+x/100)] in a bad case. The average dividend of a securities investor is τ1z[α2(P0 + C0)] + τ2[(y2/100) (11/4)Rt (1+x/100)]. The rate of return of the owner of the firm is {[(y1/100)(11/4)Rt (1+x/ 100)]  z[α2(P0 + C0)]}/[α3(P0 + C0)] in a good case and 0 in a bad case, so

112

6

Housing Policy to Supply Barrier-Free Rental Housing

that the average rate of return of the owner of the firm is τ1{[(y1/100)(11/4) Rt (1+x/100)]  z[α2(P0 + C0)]}/[α3(P0 + C0)]. (c) Suppose that 0  y2 < y1 < y*: For a securities investor, the rate of return is [(y1/100)(11/4)Rt (1+x/100)]/ [α2(P0 + C0)] in a good case and [(y2/100)(11/4)Rt (1+x/100)]/[α2(P0 + C0)] in a bad case. The average rate of return of a securities investor is τ1[(y1/100) (11/4)Rt (1+x/100)]/[α2(P0 + C0)] + τ2[(y2/100)(11/4)Rt (1+x/100)]/ [α2(P0 + C0)]. The dividend of a securities investor is, for a good case, [(y1/ 100)(11/4)Rt (1+x/100)] and, for a bad case, [(y2/100)(11/4)Rt (1+x/100)]. The average dividend of a securities investor is τ1[(y1/100)(11/4)Rt (1+x/ 100)] + τ2[(y2/100)(11/4)Rt (1+x/100)]. The rate of return of the most subordinated fund is 0 for both a good case and a bad case, which means that both the average rate of return and the dividend is 0 for the owner of the firm. Now I look at the effect of property management on the average rate of return of a securities investor. For case (a), the average rate of return of a securities investor is z% and will not change with property management. However, because y ¼ z ½α2 ðP0 þ C 0 Þ  100=½ð11=4ÞRt ð1 þ x=100Þ,

ð6:5Þ

y* will decrease by property management (which raises x). Thus, an interval in which a securities investor can receive maximum rate of return will be widened. In the case of (b) and (c), the equation of the average rate of return of a securities investor shows that it will go up by property management and resulting rise of x. These cases show that property management increases the average rate of return of a securities investor. I can also show that property management will raise the average dividend of a securities investor in the same way. In case (a), the average dividend of a securities investor is a maximum dividend of z[α2(P0 + C0)], which will not change with property management, but the interval for this maximum dividend will expand with property management. In cases (b) and (c), the average dividend of a securities investor will rise with property management. Altogether, property management will increase the average dividend of a securities investor with a rise of x. Moreover, the rate of return of a securities investor is positive on average with this scheme. In case (a), it is usual if I set the average rate of return z% is positive, since it is the upmost return. In case (b), the average rate of return is positive because 0  y2, x > 0 and z > 0. Average rate of return of a securities investor is positive in case (c), because 0  y2, y2 < y1, 0 < τ1, τ2  1 and x > 0. Therefore, there exits another effect 5. below. 5. Average rate of return of the government sector will become positive with property management. This means that the government sector will not go into budget deficit with this project.

References

113

6. The required level of barrier-free elements (that is, “bathtubs easy to step into,” “indoors without steps,” “handrails along stairs,” and “hallways wide enough for a wheelchair”) will be satisfied. 7. Minimum welfare of the nation will be satisfied by making houses safe for the elderly, reducing accidental deaths and making rental houses easier places for nursing care. 8. Nursing care expenses for both the country and individual households will be reduced. This is because making rental housing barrier-free with “indoors without steps,” “handrails along stairs,” and “hallways wide enough for a wheelchair” will enable the elderly to move by themselves without the aid of home care workers. Making housing barrier-free, the elderly will be able to take baths without home-visit bathing services, in a rental house with “bathtubs easy to step into.”

6.8

Conclusion

I proposed a policy to resolve the difficulties associated with supplying barrier-free rental housing in Japan. The plan involves a government sector becoming a securities investor in a scheme for funding rental housing. The government sector would also assist property management. In addition, low-income elderly would be offered rent subsidies. This plan entails increasing funds to construct barrier-free rental housing, increasing the rents of barrier-free housing, and making the profit rates for barrier-free housing positive, thus accelerating the supply of barrier-free rental housing. The plan would enable low-income elderly to live in barrier-free rental housing. Likelihood of accidental deaths in the rental house of elderly will be reduced and nursing care will be less difficult to deliver, which will reduce the costs of nursing care in the nation and each household.

References Cabinet Office (2019) 2019 White Paper on Aging Society, Tokyo. Foundation for Senior Citizen’s Housing (1998) Report on Research on Barrier Free Houses, Tokyo: Foundation for Senior Citizen’s Housing. Hayakawa, K. (1990) “Aging Society and Housing,” in Kanamori H. and H. Ibe (eds) Economics of Aging Society, Tokyo: University of Tokyo Press, 269–302. Hirono, K. N. (2009) “Housing Policy for the Elderly; a Policy to Build Barrier-Free Rental Housing,” Pacific Economic Review, 14(5), 694–704. Ito, A. and M. Sonoda (1994) Koureijidai wo Sumau (Live in the Stage of Old Age), Tokyo: Kenchikushiryokenkyusha. Japan Foundation for Aging and Health (2020) Health Longevity Net. https://www.tyojyu.or.jp/net/ kenkou-tyoju/tyojyushakai/sekaiichi.html

114

6

Housing Policy to Supply Barrier-Free Rental Housing

Kajima, Y. (2002) “Research on Housing Reform for the Elderly,” Zaidan News, Foundation for Senior Citizen’s Housing, 49, 26–31. Ministry of Agriculture, Forestry and Fisheries (2018) Japanese Characteristics from a Nutritional Point of View. https://www.maff.go.jp/j/keikaku/syokubunka/culture/eiyo.html Ministry of Construction Policy Research Center (1993) “The Reducing Effect of Development of Houses for the Elderly on Cost of Nursing Care,” Policy Research Center Note, 4, 1–55. Ministry of Health, Labor and Welfare (2010) Vital Statistics 2010, Tokyo. Ministry of Health, Labor and Welfare (2018a) Abridged Life Tables for Japan 2018. https://www. mhlw.go.jp/toukei/saikin/hw/life/life18/index.html Ministry of Health, Labor and Welfare (2018b) Demographic Statistics 2018. https://www.mhlw. go.jp/toukei/saikin/hw/life/life18/index.html Ministry of Land, Infrastructure, Transport and Tourism (2015) Comprehensive Survey of Housing and Living. https://www.mlit.go.jp/common/001104812.pdf National Institute of Population and Social Security Research (2019) Latest Demographic statistics 2019, Tokyo. Statistics Bureau (2018) Housing and Land Survey 2018, Tokyo. Suzuki, T., Y. Tanaka, M. Tujimura, S. Nagamachi, S. Meguro, K. Minami and E. Yamazaki (2007) “Effect of Educational Expenses on Decreasing Birthrate,” Paper reported at ISFJ Policy Forum.

Correction to: Economic Analysis of Housing Policy in Japan

Correction to: K. N. Hirono, Economic Analysis of Housing Policy in Japan, New Frontiers in Regional Science: Asian Perspectives 64, https://doi.org/10.1007/978-981-19-4925-8 Owing to an oversight on part of Springer, the original online version of the book was published with errors. The corrections listed below have been updated with this erratum. Chapter 1 Page 13, line 7 “housing market feel lack of clarity” has been changed to “housing market lack of clarity.” Page 15, line 48 In Maruyama Y. (2004) reference, Journal name has been expanded to Quarterly Journal of Housing and Land Economics. Chapter 2 Page 26, Table 2.2 footnote “*Note: Statistically different from zero at the 5% level” has been changed to “Note: * Statistically different from zero at the 5% level.”

The updated original version for this book can be found at https://doi.org/10.1007/978-981-19-4925-8 © Springer Nature Singapore Pte Ltd. 2022 K. N. Hirono, Economic Analysis of Housing Policy in Japan, New Frontiers in Regional Science: Asian Perspectives 64, https://doi.org/10.1007/978-981-19-4925-8_7

C1

C2

Correction to: Economic Analysis of Housing Policy in Japan

Chapter 3 Page 41, Line 3 of Section 3.4 “γ, and Eq. (3.6).” has been changed to “γ,―and Eq. (3.6).” Page 45, Table 3.2 footnote “*Note: Statistically different from zero at the 5% level” has been changed to “Note: * Statistically different from zero at the 5% level.” Chapter 4 Page 53, Table 4.1, column 1 “Direct Finance” and “Loan through securitization” were grouped under GHLC. Page 61, Table 4.4. column 2 The * has been removed from column 2 values. Page 61, Table 4.6, column 4 The * has been removed from column 4 values. Page 61, Table 4.7 The * has been removed from column 4 values. Chapter 5 Page 81, line 11 “(National Police Agency, 2020 [2011])” has been changed to “(National Police Agency, 2020).” Page 81, line 13 “(National Police Agency, 2020 [2011])” has been changed to “(National Police Agency, 2011).” Page 97, line 27 “(National Police Agency, 2020 [2011])” has been changed to “(National Police Agency, 2020, 2011).” Page 92, Fig. 5.5 The unit “Yen” has been inserted in X-axis. Chapter 6 Page 101, Fig. 6.1: figure and captions are updated.