공간계량모형을 이용한 갈치 어획의 공간적 상호의존성과 결정요인 분석 : 제주 해역을 중심으로 = Analyzing Spatial Interdependence and Determinants of Hairtail Catches Using Spatial Econometric Models: Evidence from the Jeju Waters

다국어 초록 (Multilingual Abstract)

This study investigates the spatial interdependence of hairtail catches and the influence of key marine environmental factors in the waters surrounding Jeju Island, where hairtail fishing is most concentrated in Korean coastal zones. As fishing grounds around Jeju have recently contracted and resource abundance has declined, quantitatively examining the complex interplay between marine environmental changes and fishing activities has become increasingly important. To address this need, we construct a quarterly high-resolution (1 arc-minute × 30 arc-second) grid panel dataset for 2020-2022 and analyze the spatio-temporal dynamics of hairtail catches. Hairtail catches serve as the dependent variable, while seawater temperature, salinity, dissolved oxygen, and chlorophyll-a represent marine environmental conditions; the number of fishing vessels and yellow croaker catches capture fishing effort and prey availability. The study area is delineated using natural breaks and Local Moran’s I (LISA) to identify High-High clusters of hairtail catches around Jeju Island. Spatial weight matrices are specified by comparing contiguity-based (Queen contiguity with first- and second-order neighbors) and distance-based (global inverse distance and inverse distance within 25 km) structures, from which the most appropriate configuration is selected. The empirical strategy first evaluates spatial autocorrelation to determine the optimal spatial weight matrix and then applies a static Spatial Durbin Model (SDM) and a dynamic SDM in sequence. The static SDM, estimated using the full-period panel data, identifies contemporaneous direct and spillover effects of individual variables, and is subsequently re-estimated on a quarterly basis to examine seasonal heterogeneity in the effect structure. The dynamic SDM, employing the same spatial specification, incorporates temporal dependence to decompose short- and long-run effects and to assess cumulative spatio-temporal impacts on hairtail catches. The results can be summarized as follows. First, natural breaks and LISA reveal statistically significant High-High clusters of hairtail catches widely distributed around Jeju Island. Global Moran’s I is positive across all spatial weight matrices and quarters, confirming strong spatial autocorrelation and the persistence of core fishing grounds. Second, the static SDM is identified as the preferred model. Salinity, vessel numbers, and yellow croaker catches exert significant positive direct effects on hairtail catches, whereas salinity and chlorophyll-a show negative indirect effects and dissolved oxygen, vessel numbers, and yellow croaker catches show positive indirect effects. These patterns suggest that both spatial substitution and spatial diffusion mechanisms are simultaneously at work. Quarterly SDM estimates further reveal pronounced seasonal heterogeneity, with salinity imposing negative direct effects in the first and fourth quarters but positive effects in the second and third quarters— patterns consistent with seasonal shifts in optimal water-mass conditions. Third, the dynamic SDM provides the best fit among dynamic specifications. Lagged own-region catches and both contemporaneous and lagged neighboring catches significantly increase current catches, offering robust evidence of cumulative spatio-temporal spillover effects driven by temporal persistence and inter-regional interactions. By integrating high-resolution spatial panel data with spatial econometric analysis, this study quantitatively elucidates the spatio-temporal interaction structure of hairtail catches and provides a unified understanding of how marine environmental conditions, fishing effort, and prey availability jointly shape the formation and seasonal evolution of fishing grounds. The combined application of static and dynamic SDMs enables systematic identification of cumulative spatio-temporal effects and seasonal heterogeneity, thereby offering econometric validation of existing oceanographic and ecological findings. Overall, the analytical framework advanced in this study provides a scientific basis for improving fishing-ground prediction and formulating resource-management and climate-change adaptation strategies in fisheries.

목차 (Table of Contents)

• Ⅰ. 서론 1
• 제1절 연구의 배경 및 목적 1
• 1. 연구 배경 1
• 2. 연구 목적 4
• 제2절 연구의 범위 및 방법 5
• 1. 연구 범위 5
• 2. 연구 방법 6
• 3. 연구 구성 8
• Ⅱ. 선행연구 검토 11
• 제1절 선행연구 동향 및 시사점 11
• 1. 갈치의 생태적 특성 관련 선행연구 11
• 2. 공간자기상관성 관련 선행연구 18
• 3. 공간계량모형 관련 선행연구 24
• 제2절 선행연구와의 차별성 32
• Ⅲ. 갈치 자원의 생태적 특성과 관리체계 34
• 제1절 생태적 특성 34
• 제2절 갈치 생산 현황 36
• 1. 우리나라 갈치 생산 현황 36
• 2. 제주지역 갈치 생산 현황 42
• 3. 소결 44
• 제3절 자원관리 제도 검토 45
• 1. TAC(총허용어획량) 제도 45
• 2. 금어기 및 금지체장 제도 47
• Ⅳ. 분석 방법 및 모형 48
• 제1절 공간가중치 행렬 48
• 제2절 공간자기상관 분석 52
• 1. 전역적 Moran’s I 검정 52
• 2. 국지적 Moran’s I 검정(LISA) 53
• 제3절 공간계량모형 56
• 1. 정적 공간계량모형(Static Spatial Models) 56
• 2. 동적 공간계량모형(Dynamic Spatial Models) 62
• 3. 공간효과 분해(직접·간접·총효과) 65
• 제4절 본 연구의 적용 논리 69
• Ⅴ. 실증 분석 71
• 제1절 연구 범위의 공간적 한정 71
• 1. 분석 대상 및 공간 단위 71
• 2. 어획량의 공간적 분포 및 군집 특성 분석 73
• 3. 최종 분석 범위 설정 84
• 제2절 변수 정의 및 자료 구성 86
• 1. 독립변수의 정의 및 구성 86
• 2. 데이터 처리 및 패널자료 구축 90
• 제3절 공간자기상관 분석 및 최적 행렬 선정 92
• 1. 공간가중치 행렬의 구성 92
• 2. 전역적 Moran’s I 검정 결과 95
• 3. 공간가중치 행렬의 적합성 평가 및 선정 98
• 제4절 정적 공간패널모형 분석 100
• 1. 분석자료의 기초통계량 100
• 2. 모형 선정 및 적합성 검정 103
• 3. 정적 SDM 모형 추정 및 효과 분해 111
• 제5절 동적 공간패널모형 분석 123
• 1. 분석자료의 기초통계량 123
• 2. 모형 적합성 검정 및 선정 128
• 3. 장·단기 효과 분해와 강건성 검토 132
• Ⅵ. 결론 145
• 제1절 연구 및 분석 결과 요약 145
• 1. 연구 요약 145
• 2. 분석 결과 요약 146
• 제2절 연구의 의의 및 정책적 시사점 149
• 1. 연구의 의의 149
• 2. 정책적 시사점 152
• 제3절 연구의 한계 및 향후 연구과제 155