氏名　NGUYEN CAO Y
学位の種類　博士（工学）
学位記番号　博甲第634号
学位授与の日付　平成24年8月31日
学位論文題目　Firm Re-location Pattern and Location Choice Model Incorporating Spatial Effects　（空間効果を考慮した事業所立地選択モデル）
論文審査委員
　主査　准教授　佐野　可寸志
　副査　教授　中出　文平
　副査　准教授　樋口　秀
　副査　名誉教授　松本　昌二
　副査　東京海洋大学　教授　兵藤　哲朗

• 論文目次
• 論文要旨
• 論文審査の結果の要旨

［平成24（2012）年度博士論文題名一覧］ ［博士論文題名一覧］に戻る．

Contents　page
Chapter 1:　INTRODUCTION　p.1
　1.1　INTRODUCTION　p.1
　1.2　RESEARCH PROBLEM　p.2
　1.2.1　Defining Firm Location Choice Decision Process　p.3
　1.2.2　Defining Firm Re-Location Process　p.4
　1.3　RESEARCH OBJECTIVES　p.5
　1.4　RESEARCH SCOPE　p.5
　1.5　RESEARCH ORGANIZATION　p.6
Chapter 2:　LITERATURE REVIEW　p.10
　2.1　INTRODUCTION　p.10
　2.2　LOCATION CHOICE MODELS　p.10
　2.2.1　Firm Location Choice Models　p.10
　2.2.2　Discrete Choice Models with Structuralized Spatial Effects　p.11
　2.3　FIRM LOCATION CHOICE MODELS　p.13
　2.3.1　Multinomical Logit Approach　p.13
　2.3.2　Nested Logit Model Approach　p.13
　2.3.3　A Micro Simulation Model Approach　p.14
　2.3.4　An Ordered Logit Approach　p.15
　2.3.5　Cross-section Regression Approach　p.15
　2.3.6　Gravity Type Approach　p.15
Chapter 3:　CONCEPTUAL FRAMEWORK　p.16
　3.1　INTRODUCTION　p.16
　3.2　CONCEPTUAL FRAMEWORK OF LOCATION CHOICE MODELS　p.17
　3.3　CONCEPTUAL FRAMEWORK OF RE-LOCATION MODELS　p.20
Chapter 4:　LOCATION CHOICE AND RE-LOCATION BEHAVIOR: SOME FACTS　p.24
　4.1　INTRODUCTION　p.24
　4.2　STUDY AREA　p.24
　4.3　SURVEY DATA　p.25
　4.3.1　Tokyo Metropolitan Goods Movement Survey（TMGMS 2004）　p.25
　4.3.2　Tokyo Metropolitan Goods Movement Survey（TMGMS 1994）　p.26
　4.3.3　Population Census of Japan　p.26
　4.3.4　Land Price Survey of Japan　p.27
　4.3.5　Establishment and Enterprise Census（EEC）（1999）　p.27
　4.3.6　Road Traffic Census（RTC）（MLIT 1999）　p.28
　4.4　FIRM LOCATION CHOICE BEHAVIOR AND SOME FACTS　p.28
　4.4.1　The Relationship between Number of Firms and some Influence Factors　p.33
　4.5　FIRM LOCATION DECISION BEHAVIOR AND SOME FACTS　p.42
　4.5.1　Move/Stay Decision and Some Facts　p.43
　4.5.2　Some Factors in Zone Choice Decision　p.53
　4.6　CONCLUSIONS　p.58
Chapter 5:　LOCATION CHOICE MODELS INCORPORATING SPATIAL EFFECTS　p.60
　5.1　INTRODUCTION　p.60
　5.2　STUDY METHODOLOGY　p.62
　5.2.1　Mixed Logit Models　p.62
　5.2.2　Modeling Spatial Effects　p.63
　5.2.3　Mixed Logit Models for Location Choice Decision Behavior　p.64
　5.2.4　Spatial Correlation among Firms in Deterministic Term　p.65
　5.2.5　Spatial Correlation among Zones in the Error Term　p.66
　5.2.6　Simulation　p.69
　5.3　DATA COLLECTION　p.72
　5.4　RESULTS AND DISUCUSSION　p.73
　5.4.1　The Estimation Results of the Proposed Model by MNL　p.73
　5.4.2　The Estimation Results of the Proposed Model by MNLS　p.77
　5.4.3　The Estimation Results of the Proposed Model byMLS　p.82
　5.4.4　The Comparison of Estimation Results for Each Kind of Industry　p.89
　5.5　CONCLUSIONS　p.94
Chapter 6:　FIRM RE-LOCATION MODELS INCORPORATING SPATIAL EFFECTS　p.96
　6.1　INTRODUCTION　p.96
　6.2　STUDY METHODOLOGY　p.97
　6.2.1　Move/Stay Decision Process　p.99
　6.2.2　Zone ChoiceDecision Process　p.100
　6.2.3　Re-location Models by Nested Logit Model　p.101
　6.3　DATA COLLECTION　p.104
　6.4　RESULTS AND DISUCUSSION OF THE MOVE/STAY DECISION　p.105
　6.4.1　The Move/stay Decision for Chemical Manufactures　p.105
　6.4.2　The Move/stay Decision for Machineryl Manufactures　p.107
　6.4.3　The Move/stay Decision for Retailers　p.109
　6.4.4　The Comparison of Estimation Results for Move/stay Decision　p.111
　6.5　RESULTS AND DISCUSSION OF THE ZONE CHOICE DESION　p.116
　6.5.1　The Comparison of Estimation Results for Zone Choice Decision　p.116
　6.5.2　TheResults of Firm Re-location Decision by Neted Logit Model　p.117
　6.6　CONCLUSIONS　p.121
Chapter 7:　CONCLUSIONS AND RECOMMNDATIONS　p.122
　7.1　CONCLUSIONS　p.122
　7.1.1　Location Choice Model Incorporating Spatial Effects　p.122
　7.1.2　Firm Re-location Models Incorporating Spatial Interactions　p.123
　7.2　RECOMMENDATIONS　p.124
REFERENCES　p.126
APPENDICES　p.140

In order to survive in an increasingly competitive business environment, individual firms must constantly adapt themselves to new and changing circumstances. Relocation of a firm, therefore, can be considered as a form of the adaptation. Besides, understanding the factors that play a key role in the decisions that businesses make on where to locate is very important in order to determine the most effective mechanisms to attract firms, and thereby maintain a healthy economy. In urban economics, the related individual firms which gather together will gain benefits such as lower cost of production and a larger market share that one firm can achieve. A firm, while making a relocation decision or location choice decision, does not act in isolation; in contrast, firms are influenced by others who located nearby. Therefore, the main purpose of the dissertation is to develop a location choice decision model and relocation model of each individual firm under considering the spatial effects in urban areas. Hence, the main contribution of this research analyses the influence of these factors on location choice and relocation decision behavior of an individual firm in order to better understanding the key factors. In addition, how the spatial effect on the firm location choice and relocation behaviors should be considered carefully in the research process.

This study presents an overview of a location choice model for individual firms and this model takes into account spatial effects. A modeling framework is developed to analyze decisions regarding location choice for individual firms using series of discrete choice model including multinomial logit model and mixed logit models. In this framework, spatial effects such as the spatial correlation among firms in deterministic terms and the spatial correlation among zones in the error term are captured by mixed logit models. In a case study, the developed framework is applied to the Tokyo metropolitan area. The model is tested for retailers, wholesalers, warehouses and manufacturers.

This dissertation develops a relocation model to analyze decisions regarding the relocation process for individual firms by using discrete choice models. The relocation decision process of each individual firm which decides to move and relocate to the other zone is the product of the marginal probability of this individual firm that will move and the conditional probability of firm that chooses a location. In this framework, two decision points in the relocation process are assumed and maintained in the micro-simulation modeling. The first decision, move or non move, is modeled by using a binary logit form with outcome the probability of moving. The second decision, choosing the destination location, is modeled by a mixed logit model incorporating spatial effects with the outcome of the conditional probability of choosing a zone.

The result has highlighted the influence of spatial effects on the decision behavior of individual firms for location choice. A mixed logit model is applied and the results indicate that the location choice decision of firms depends on the number of employees in a particular firm, transportation cost of each firm and the attributes of the zone such as the size of the population and the number of employees in the zone. Moreover, the results suggest that individual firms prefer to locate in zones that have a higher population density and a higher number of employees. Furthermore, the study clarifies the factors that play a key role in the decisions made by individual companies in selecting a location. Our findings confirm the important role of structuralized spatial effects and zone attributes in the decision-making process of individual firms.

For the move/stay decision of each individual firm, the study found that manufacturer firms with high number of employee and large floor area have a lower probability of relocation based on the negative sign of the estimated parameters. In addition, the high land price of the current location increases the need to move for all manufacturers and retailers. In a similar way, the distance to the nearest IC highway has a positive effect on their moving decision. In other words, chemical manufacturers, and machinery manufacturers prefer to be located near highway onramps. Furthermore, the individual firms which have a high demand to get a large land, need to be near to customers or major customers, need to be near IC highway, and need to be convenient for employees to commute, also have a high mobility to move. The pressures of many problems such as loading and unloading at road in front of firm, parking space, can contribute to increase the level of moving decision for chemical manufacturers, machinery manufacturers and retailers. Last but not least of move/stay decision model, the log likelihood ratio and AIC values of the proposed model for move/stay decision of each individual firm being rather high indicate that the model structure can well explain the moving decision of individual firm relocation behavior.

Regarding the zone choice decision, the study found that the attributes of zones such as population density of each zone, average land price and number of employees of zone are the important attractive factors which have an influence on the zone choice decision of each individual firm. This result is similar for chemical manufacturers, machinery manufacturers and retailers. In general, the individual firms prefer to move to the zones which have a short distance from the current location. The results also confirm again the important role of spatial interactions in the firm relocation decision process. The relocation decision structure has been tested by nested logit model. The results of the nested logit model indicate that the model has an acceptable value of the log-likelihood ratio and AIC values. However, some limitations of Nested logit model suggest that the relocation hierarchical structure can be replaced with the other model structure which is likely to be proper. The results suggest that the relocation process which includes two decisions move/stay decision and zone choice decision is appropriate in the proposed firm relocation model in this dissertation.

The results of this research still open for the future research which wants to study on the individual firm relocation models. The obtained results still leaves ample room for improvement because the present study is limited to a small data set for the estimation and common structures. It is expected that the better model performance will be achieved with an improved, larger set of data and specific structures.

　本論文は，「Firm Re-location Pattern and Location Choice Model Incorporating Spatial Effects（空間効果を考慮した事業所立地選択モデル）」と題し，以下の7章より構成されている．

　第1章「Introduction」では，統合土地利用交通モデルにおける事業所立地選択モデルの持つ問題点について述べ，本研究の必要性と適用範囲を記述している．
　第2章「Literature review」では，離散選択モデルを用いた，既往の事業所立地選択モデルや事業所再立地モデルの既存の研究についてまとめている．
　第3章「Conceptual Framework」では，ミックスドロジットモデルを用い，空間相互作用を考慮した立地選択離散選択モデルモデルのフレームワークを記述している．
　第4章「Location Choice Decisions and Re-location Decisions: Some Facts」では，東京都市圏物資流動調査を用い，エリアや業種の違いによる立地企業の特性や，個々の企業の立地選択要因の抽出を行っている．
　第5章「Location Choice Models for Individual Firms with Consideration of Spatial Effects」では，空間相互作用を考慮した複数の立地選択モデルの定式化と，実データを用いたモデルの推定を行い，各モデルの特徴をまとめている．
　第6章「Firm Re-location Models incorporating Spatial Interactions」では，立地選択モデルの上位モデルとなる再立地モデルの定式化と，実データを用いたモデルの推定を行い，その特徴をまとめている．
　第7章「Conclusions and Recommendations」では，各章で示したモデルの特徴をまとめるとともに，今後の課題をまとめている．

　以上のように，本論文は，空間相互作用を考慮した立地選択モデルを構築することにより，新しい知見を与えている．よって，本論文は工学上及び工業上貢献するところが大きく，博士（工学）の学位論文として十分な価値を有するものと認める．