Rail Stations and Housing Price Capitalization : Express, Heavy, and Light Rail in the Seoul Metropolitan Area = 철도역과 주택가격 자본화: 수도권 광역급행철도·중전철·경전철 비교

다국어 초록 (Multilingual Abstract)

Large rail investments can reshape commuting conditions and land values, yet the magnitude and timing of housing-price capitalization depend critically on service design (speed, stopping pattern, and through-running), a station’s network role (transfer opportunities and direct connections), and implementation credibility under uncertainty. This dissertation investigates how rail-induced changes in job-center accessibility are capitalized into apartment prices in the Seoul Metropolitan Area (SMA), with a particular focus on the Great Train eXpress (GTX) and explicit comparisons to conventional heavy rail transit (HRT) and light rail transit (LRT). The study assembles a comprehensive transaction-level dataset covering the universe of SMA apartment sales from January 2006 to December 2024 (about 4.6 million observations), deflates prices to real December 2024 values, and links each transaction to apartment-complex attributes and neighborhood accessibility measures based on road-network shortest-path distances. Station areas are defined as properties within 500 meters (network distance) of the nearest rail station. To identify causal effects and their evolution over time, the dissertation codes a harmonized project-phase structure—announcement, construction, and opening—and applies event-phase difference-in-differences (DID) and triple-difference (DDD) designs to recover phase-specific station-area premia and spatial heterogeneity across Seoul versus non-Seoul jurisdictions and within-corridor segments.

Results show that GTX capitalization is substantially larger and forms earlier than capitalization associated with conventional metro and LRT projects. For GTX-A, cumulative station-area premia rise sharply at the announcement stage (about 35%), increase further during construction (about 46%), and remain large after the initial opening (about 43%), whereas the pooled LRT premium after opening is modest (about 5%) and the heavy-rail benchmark line exhibits limited or even negative premia. More broadly, capitalization begins well before service starts, consistent with forward-looking valuation: announcement effects reflect expectation formation, construction tends to amplify effects as perceived deliverability improves, and opening primarily validates and adjusts earlier premia rather than initiating them. For pooled GTX, Seoul–non-Seoul DDD estimates indicate that announcement-stage station-area premia are small and slightly negative (about −2% in Seoul and −5% in non-Seoul), while construction-stage premia are similar (about +12% in Seoul and +13% in non-Seoul).

Within GTX, capitalization is highly uneven across corridors and directions: GTX-A’s southern segment shows large premia from announcement to opening (announcement +49%, construction +41%, opening +37%), whereas the northern segment responds little at announcement (about 0%) and only shows meaningful catch-up after construction (+24% at construction and +23% at opening). GTX-C shows moderate premia in the southern segment (+17% at announcement and +17% at construction) but delay‑ and uncertainty‑linked discounts in the northern segment (−26% at announcement and −18% at construction). GTX-B (west vs. east) exhibits small discounts in the western segment (−5% at announcement and −4% at construction) while the eastern segment is near zero at both stages. These findings imply that capitalization in a polycentric metropolis is governed not only by proximity to rail infrastructure but by the quality of job-center access improvements (CBD/GBD/YBD orientation), network connectivity, and evolving project credibility. The dissertation highlights the need for appraisal and policy frameworks that account for anticipatory market responses and for complementary land-use, housing-supply, and feeder-network strategies to manage distributional pressures around major rail investments.

국문 초록 (Abstract)

대규모 철도 투자는 통근 여건과 토지가치를 재편할 수 있지만, 주택가격 자본화의 크기와 발생 시점은 철도 서비스 설계(속도·정차 패턴·도심 관통 여부), 네트워크 내 역의 역할(환승·직결성), 그리고 사업 진행의 신뢰성(지연·불확실성)에 따라 크게 달라질 수 있다. 본 연구는 수도권 광역급행철도(GTX)를 중심으로 중전철(도시철도) 및 경전철(LRT) 사업을 함께 비교하여, 철도 유형별·단계별로 고용 중심지 접근성 개선이 아파트 실거래가격에 어떻게 자본화되는지를 규명한다. 분석 자료는 2006년 1월~2024년 12월 수도권 아파트 실거래 전수(약 460만 건)이며, 단지 특성과 주변 시설 접근성(도로망 최단경로 거리)을 결합하고, 역세권은 역까지 네트워크 거리 500 m 이내로 정의하였다. 방법론적으로는 발표–착공–개통의 공통 사업 단계 체계를 코딩한 뒤, 사건시점 이중차분(DID)과 삼중차분(DDD)을 적용하여 단계별 효과와 서울/비서울, 노선 내 남·북/동·서 등 공간적 이질성을 식별하였다.

분석 결과, GTX의 자본화 효과는 기존 도시철도 및 경전철보다 크고, 더 이른 시점부터 형성되는 경향이 확인된다. 특히 GTX‑A의 역세권 누적 프리미엄은 발표 이후 약 35%, 착공기 약 46%, (부분)개통 이후에도 약 43% 수준으로 추정되는 반면, 경전철 평균 효과는 개통 이후 약 5% 내외에 그치고, 중전철(9호선)은 유의한 양(+)의 프리미엄이 제한적이거나 음(–)의 패턴도 관찰된다. 또한 자본화는 개통 후에만 나타나는 것이 아니라 발표 단계의 기대효과로 시작해 착공 과정에서 사업 신뢰성 갱신에 따라 확대되며, 개통은 기존 기대를 검증·조정하는 역할을 하는 것으로 해석된다. 또한 GTX 전체를 대상으로 한 서울–비서울 DDD 결과, 발표(고시) 단계 역세권 프리미엄은 서울 약 −2%·비서울 약 −5%로 크지 않으나, 착공기에는 서울 약 +12%와 비서울 약 +13%로 유사하게 추정되었다.

한편 동일한 GTX 내에서도 방향·구간별 이질성이 크게 나타났다. GTX‑A는 남부 구간에서 발표 단계부터 역세권 프리미엄이 크게 관측된 반면(발표 +49%, 착공 +41%, 개통 +37%), 북부 구간은 발표 단계 반응이 0% 내외로 미미하고 착공 이후 프리미엄이 형성되는 ‘후행’ 양상을 보였다(착공 +24%, 개통 +23%). GTX‑C는 남부 구간에서 프리미엄이 나타난 반면(발표 +17%, 착공 +17%), 북부 구간은 지연·불확실성과 결합된 할인(음의 자본화)으로 추정되었다(발표 −26%, 착공 −18%). GTX‑B는(본 모형에서 서·동 구간으로 구분) 서부 구간에서 소폭 할인(발표 −5%, 착공 −4%), 동부 구간은 0% 내외로 추정되어, 동일 노선 내부에서도 기대 형성과 사업 진척의 차이가 가격 반응을 좌우함을 시사한다. 이러한 결과는 철도 자본화가 단순한 역세권 여부나 중심성만으로 설명되기보다, 주요 고용 중심지(CBD·GBD·YBD)로의 접근성 개선의 ‘질’, 네트워크 직결성, 그리고 사업의 신뢰성이 함께 작동함을 시사한다. 정책적으로는 급행형 광역철도에서 기대·신뢰성에 따른 선(先)자본화가 크게 발생할 수 있으므로, 평가와 관리가 개통 이후뿐 아니라 발표–착공 단계까지 확장되어야 하며, 역세권 주거시장 압력에 대응하기 위한 토지이용·공급·연계교통 등 보완 정책이 요구된다.

목차 (Table of Contents)

• Table of Contents
• Chapter 1. Introduction 1
• 1.1 The broader context of rail and accessibility 1
• 1.2 Research objectives and questions 4
• Chapter 2. Literature Review 7
• 2.1 Conceptual foundation: job-center accessibility and housing-price capitalization in polycentric metros 7
• 2.2 GTX as a job-accessibility shock and expected housing-price capitalization 13
• 2.3 International analogues of metropolitan express, through-running systems 19
• 2.3.1 London, UK — Elizabeth Line (formerly “Crossrail”) 19
• 2.3.2 Paris, France — RER (Réseau Express Régional; Regional Express Network) 21
• 2.3.3 Paris, France — Grand Paris Express (GPE) 23
• 2.3.4 Moscow, Russia — МЦД (Московские центральные диаметры; Moscow Central Diameters, MCD) 25
• 2.3.5 Tokyo, Japan — つくばエクスプレス (Tsukuba Express, TX) 28
• 2.4 Empirical literature in Korea: accessibility changes and price capitalization 31
• 2.4.1 Common approaches, data, and measurement in early GTX-related studies 31
• 2.4.2 GTX-focused capitalization evidence across milestones and distance gradients 32
• 2.4.3 Accessibility-focused analyses and simulations of GTX impacts 35
• 2.4.4 Synthesis, limitations, and remaining gaps in domestic evidence 37
• 2.5 International empirical literature: express/through-running rail, accessibility gains, and capitalization 39
• 2.5.1 London, UK — Elizabeth Line (formerly “Crossrail”) 39
• 2.5.2 Paris, France — RER (Réseau Express Régional; Regional Express Network) 43
• 2.5.3 Paris, France — Grand Paris Express (GPE) 46
• 2.5.4 Moscow, Russia — МЦД (Moscow Central Diameters, MCD) 50
• 2.5.5 Tokyo, Japan — つくばエクスプレス (Tsukuba Express, TX) 52
• 2.6 Integrated synthesis: gaps and how this dissertation identifies incremental GTX effects 55
• Chapter 3. Theoretical Framework and Hypotheses 59
• 3.1 Conceptual framework: rail investments, job-center accessibility, and capitalization 59
• 3.2 Expectations, uncertainty, and project stages 62
• 3.3 Hypotheses 64
• 3.3.1 Rail-type and line-level station-area capitalization (H1) 64
• 3.3.2 Network integration, transfer hubs, and station roles (H2) 66
• 3.3.3 Event-stage dynamics of capitalization (H3) 66
• 3.3.4 Core–periphery heterogeneity and distributional outcomes (H4) 67
• 3.3.5 Corridor and job-center heterogeneity (H5) 68
• 3.3.6 Placebo dynamics and identification (H6) 70
• Chapter 4. Data Collection and Methodology 71
• 4.1 Data sources and study area 71
• 4.2 Descriptive statistics and correlations 80
• 4.3 Period and Event Definitions 85
• 4.4 Identification strategy and models 90
• 4.4.1 DID model: Event phase × station-area treatment 91
• 4.4.2 DDD model: Event phase × station-area treatment × subarea 92
• Chapter 5. Results 94
• 5.1 DID Analysis: Line-Level Announcement–Construction–Opening Effects (Event × Station Area) 94
• 5.2 Triple-Difference (DDD): Seoul vs. Non-Seoul Effects by GTX Line 106
• 5.3 Triple-Difference (DDD): East–West and North–South Heterogeneity of GTX Lines 111
• 5.3.1 GTX-A: Strong capitalization in the south and delayed catch-up in the north 111
• 5.3.2 GTX-C: Stable premia in the south versus discounts in the north 112
• 5.3.3 GTX-B: Weak heterogeneity and small station-area effects overall 113
• 5.3.4 Summary: Direction, CBD accessibility, and capitalization 114
• 5.4 Synthesis of Results and Hypothesis Evaluation (H1–H6) 117
• 5.4.1 Rail type, speed, and corridor centrality (H1) 117
• 5.4.2 Network integration and transfer hubs (H2) 118
• 5.4.3 Event-stage timing dynamics of capitalization (H3) 119
• 5.4.4 Seoul vs. non-Seoul disparities and distributional outcomes (H4) 120
• 5.4.5 Corridor direction and job-center orientation (H5) 121
• 5.4.6 Placebo dynamics and implementation uncertainty (H6) 123
• 5.4.7 Summary across hypotheses 123
• Chapter 6. Discussion 126
• 6.1 What makes GTX capitalization systematically larger 126
• 6.2 Why transfer hubs matter less for GTX and more for LRT 127
• 6.3 Expectations, uncertainty, and why construction is often the inflection point 128
• 6.4 Polycentric employment accessibility as the organizing principle of GTX premia 130
• 6.5 Distributional outcomes are dynamic: early peripheral gains, later recentering 132
• 6.6 Corridor direction, segment completion, and the meaning of “partial opening” 133
• Chapter 7. Conclusion 136
• 7.1 Summary of the study and key findings 136
• 7.2 Academic and methodological contributions 138
• 7.3 Policy implications 139
• 7.4 Limitations and directions for future research 140
• References 142
• Appendix 152
• Abstract (Korean) 166
• List of Tables
• Table 1. GTX vs. Conventional Commuter/Metro Rail: Design & Operations and Implications for Price Capitalization 12
• Table 2. Network, Service, and Speed Comparison: GTX, Elizabeth Line, RER, GPE, MCD, and TX 30
• Table 3. Near-Station Premium by Distance: Point Estimates with 95% Confidence Intervals 75
• Table 4. Variable Type 78
• Table 5. Descriptive Statistics and Correlation 81
• Table 6. Project Phases by Line (Announcement, Construction, and Opening) 89
• Table 7. DID Estimates by Line and Milestone (Focus: Event × Station-Area Treatment) 102
• Table 8. Phase-Specific Cumulative Station-Area Price Premia by Line (Percent, Converted from Table 7) 106
• Table 9. Triple-Difference (DDD) Estimates of GTX Station Premiums: Seoul vs. Non-Seoul 110
• Table 10. Triple-Difference (DDD) Estimates of GTX Station Premiums by Corridor: South vs. North and West vs. East 116
• List of Figures
• Figure 1. Map of the GTX-A Line 15
• Figure 2. Map of the GTX-B Line 16
• Figure 3. Map of the GTX-C Line 17
• Figure 4. Map of the GTX-A/B/C Lines 18
• Figure 5. London Metro Map 20
• Figure 6. Elizabeth Line Map 20
• Figure 7. Paris RER Map 22
• Figure 8. Grand Paris Express Map 24
• Figure 9. Moscow Metro Map 26
• Figure 10. Moscow Central Diameters (MCD) Map 27
• Figure 11. Tsukuba Express (TX) Map 29
• Figure 12. Research Framework for Rail-Transit Capitalization (Variables,Temporal and Regional Moderators, and Outcome). 60
• Figure 13. Study Area 72
• Figure 14. Distribution of Apartment Prices Before and After Log Transformation (Deflated to Dec 2024) 74
• Figure 15. Near-Station Premium by Distance: Point Estimates with 95% Confidence Intervals 75
• Figure 16. Overlap of GTX, HRT, and LRT Station Areas and Allocation of Treated and Control Groups 76
• Figure 17. Schematic Illustration of Registry-Based Negative Floor Numbering (e.g., −4) in Sloped-Site Apartment Buildings. 83
• Figure 18. Approval and Delivery Process for Metropolitan Rail (GTX): Mapping of Study Events (Announcement = Basic Plan Notification; Construction = Groundbreaking; Operation = Opening) 87