네트워크 분석, 엔트로피 및 복잡성 지표 기반 국내 반도체 산업의 구조 변화 분석

In this study, we empirically analyze the structure of the Korean semiconductor industry by employing entropy-based indicators (Shannon entropy, Gini coefficient, Herfindahl–Hirschman Index), the Economic Complexity Index (ECI), and network analysis. Two structural changes in the semiconductor industry were identified: (1) the temporty shift towrds integration in 2020, and (2) the reversal of complexity trends among regions in 2023–2024. At the national level, firms related to the semiconductor sector and their positions within the semiconductor value chain were classified using company websites and descriptive materials. Ultimately, 2,957 firms were categorized into sectors. Among these, transaction data could be extracted for 1,212 firms, yielding 66,210 firm-to-firm transaction records. The transaction network exhibits characteristics of scale-freeness, small-world properties, modularity, and hierarchy, forming an integrated structure centered on a small number of core firms. Between 2018 and 2024, the industry demonstrated a gradual trend toward specialization, although a temporary shift toward integration was observed in 2020. The materials, components, and equipment sectors showed relatively high levels of specialization, whereas the front-end processing and design sectors remained more integrated. At the regional level, Gyeonggi, Chungnam, and Chungbuk displayed high levels of integration, while the ECI values were higher in Gyeongnam and Busan. Notably, the rise in ECI for Gyeongnam, Busan, and Gwangju, alongside the relative decline of Gyeonggi, Chungnam, and Chungbuk, became evident in 2023–2024, suggesting the need for in-depth examination of government initiatives.