學術產出-Theses
Article View/Open
Publication Export
-
題名 台灣半導體供應鏈：網絡分析
Taiwan Semiconductor Supply Chain: A Network Analysis作者 陳維勛
Chen, Wei-Xun貢獻者 何靜嫺
Ho, Shirley J
陳維勛
Chen, Wei-Xun關鍵詞 網絡分析
冪次分配
中心性
集團
專利數量
Networks
Power Law
Degrees of Centrality
Cliques
Patent Numbers日期 2023 上傳時間 2-Aug-2023 13:42:56 (UTC+8) 摘要 台灣半導體產業在世界上佔重要的地位。在台灣半導體供應鏈上游、中游和下游廠商包含生產超過27種產品。原本傳統的計算方式像是market share或是HHI無法完全的描述整個半導體供應鏈。所以我們用廠商營收的相關性建立三種不同的網路(threshold, MST and PMFG networks)，在檢查這三個網路是否具有實證網路的特徵。我們的研究表明PMFG networks可以最好的描述網路因為網路degree服從power law distribution。使用建構出的PMFG網路計算描述網路和廠商的統計量，包含中心性、加權中心性和cliques。並且利用clique分析進一步將節點依照“production-related” 或 ‘product-related”進行分類，接著利用中心性和專利數量的關係，看是否創新行為是否具有傳染效果。我們的研究發現(1) "production-related" cliques，當cliques至少包含上、中、下游其中一家廠商他的營運成本會比較低。 (2)比較 “cliques producing multiple products” 和 “cliques producing few products” 發現生產比較多產品的cliques他的EPS會比生產比較少產品的cliques來得高。(3)將cliques與金融上的表現連結發現當“three-categories cliques” 裡的廠商股價的平均相關係數最高。 (4)我們研究中心性和廠商專利數量之間的關係，betweenness centrality (BC), closeness centrality (CC), and eigenvector centrality (EC)對於專利數量有正且顯著的影響。 (5) 計算加權的中心性研究加權中心性與專利數量的關係。發現每個節點都使用他的總資產進行加權weighted closeness centrality對於專利數量的影響最大。換句話來說如果每個節點都使用廠商大小來加權我們可以進一步找出影響創新的網路特徵。
Taiwan’s semiconductor industry plays a crucial role in the globe. Our firms cover all upstream, midstream and downstream of the manufacturing process, and provide more than 27 products. Traditional metrics, such as market share or HHI, cannot fully describe the structure of semiconductor industry. As there is lack of actual transactions database, we employ methods that use firms’ revenue correlations to construct three networks (threshold, MST and PMFG networks), then we examine whether the three networks conform to the empirical characteristics of networks. Our analyses suggest that the PMFG network can best describe our data and follow the power law distribution.We then calculate several metrics to measure the network properties and firm-specific characteristics in the constructed PMFG network, including particularly the degrees of centrality and the node-weighted degrees of centrality, and “cliques”. To further investigate the properties of cliques, we classify cliques according to whether the nodes are “production-related” or they are ‘product-related”. Finally, we investigate the relationship between firms’ degrees of centrality and the number of patents they hold, to see if there are contagious effects on firms` innovation activities.Our results show that (1) for "production-related" cliques, if a clique contains at least one firm each from the upstream, downstream, and midstream, then the operating costs tend to be lower. (2) We compare “cliques producing multiple products” to “cliques producing few products”, and found that cliques producing multiple product categories tend to have higher average EPS than those producing fewer categories. (3) We examined the financial connection within cliques, and found that the “three-categories cliques” have the highest average correlation coefficient in member firms’ stock prices. (4) We explored the relationship between the centrality measures of firms and the number of patents, and found that betweenness centrality (BC), closeness centrality (CC), and eigenvector centrality (EC) had a positive and significant effect on patent counts. (5) We calculated the asset-weighted centrality to examine its impact on the number of patents. We found that after each node is weighted by the proportion of total asset, the weighted closeness centrality becomes most relevant for innovations. In other words, after weighing each node by its relative firm size, we can further identify the most relative network feature for innovations.參考文獻 Alstott, J., Bullmore, E., & Plenz, D. (2014). powerlaw: a Python package for analysis of heavy-tailed distributions. PloS one, 9(1), e85777.Arora, A., Ceccagnoli, M., & Cohen, W. M. (2008). R&D and the patent premium. International journal of industrial organization, 26(5), 1153-1179.Audretsch, D. B., & Acs, Z. J. (1991). Innovation and size at the firm level. Southern Economic Journal, 739-744.Birch, J., Pantelous, A. A., & Soramäki, K. (2016). Analysis of correlation based networks representing DAX 30 stock price returns. Computational Economics, 47(4), 501-525.Chen, X., Shangguan, W., Liu, Y., & Wang, S. (2021). Can network structure predict cross-sectional stock returns? Evidence from co-attention networks in China. Finance Research Letters, 38, 101422.Chi, K. T., Liu, J., & Lau, F. C. (2010). A network perspective of the stock market. Journal of Empirical Finance, 17(4), 659-667.Eryiğit, M., & Eryiğit, R. (2009). Network structure of cross-correlations among the world market indices. Physica A: Statistical Mechanics and its Applications, 388(17), 3551-3562.Falasca, M., Zobel, C. W., & Cook, D. (2008, May). A decision support framework to assess supply chain resilience.Hearnshaw, E. J., & Wilson, M. M. (2013). A complex network approach to supply chain network theory. International Journal of Operations & Production Management.Hong, M. Y., & Yoon, J. W. (2022). The impact of COVID-19 on cryptocurrency markets: A network analysis based on mutual information. Plos one, 17(2), e0259869.Huang, W. Q., Zhuang, X. T., & Yao, S. (2009). A network analysis of the Chinese stock market. Physica A: Statistical Mechanics and its Applications, 388(14), 2956-2964.Kenett, D. Y., Tumminello, M., Madi, A., Gur-Gershgoren, G., Mantegna, R. N., & Ben-Jacob, E. (2010). Dominating clasp of the financial sector revealed by partial correlation analysis of the stock market. PloS one, 5(12), e15032.Kim, J., Lee, S. J., & Marschke, G. (2009). Relation of firm size to R&D productivity.Li, Y., Jiang, X. F., Tian, Y., Li, S. P., & Zheng, B. (2019). Portfolio optimization based on network topology. Physica A: Statistical Mechanics and its Applications, 515, 671-681Nerkar, A., & Roberts, P. W. (2004). Technological and product‐market experience and the success of new product introductions in the pharmaceutical industry. Strategic Management Journal, 25(8‐9), 779-799.Nobi, A., Maeng, S. E., Ha, G. G., & Lee, J. W. (2013). Network topologies of financial market during the global financial crisis. arXiv preprint arXiv:1307.6974.Onnela, J. P., Kaski, K., & Kertész, J. (2004). Clustering and information in correlation based financial networks. The European Physical Journal B, 38(2), 353-362.Pakes, A., & Griliches, Z. (1980). Patents and R&D at the firm level: A first report. Economics letters, 5(4), 377-381.POSFAI, M., & BARABASI, A. L. (2016). Network science. Cambridge University Press.S. Wasserman, K. Faust, Social Network Analysis(1994).Cambridge University Press, CambridgeSingh, A., Singh, R. R., & Iyengar, S. R. S. (2020). Node-weighted centrality: a new way of centrality hybridization. Computational Social Networks, 7(1), 1-33.Tumminello, M., Aste, T., Di Matteo, T., & Mantegna, R. N. (2005). A tool for filtering information in complex systems. Proceedings of the National Academy of Sciences, 102(30), 10421-10426.Tumminello, M., Di Matteo, T., Aste, T., & Mantegna, R. N. (2007). Correlation based networks of equity returns sampled at different time horizons. The European Physical Journal B, 55(2), 209-217.White, E. P., Enquist, B. J., & Green, J. L. (2008). On estimating the exponent of power‐law frequency distributions. Ecology, 89(4), 905-912.Wu, L. (2015). Centrality of the supply chain network. Available at SSRN 2651786.Zhang, Y. (2009). Stock market network topology analysis based on a minimum spanning tree approach (Doctoral dissertation, Bowling Green State University).Zhao, L., Wang, G. J., Wang, M., Bao, W., Li, W., & Stanley, H. E. (2018). Stock market as temporal network. Physica A: Statistical Mechanics and its Applications, 506, 1104-1112. 描述 碩士
國立政治大學
經濟學系
110258025資料來源 http://thesis.lib.nccu.edu.tw/record/#G0110258025 資料類型 thesis dc.contributor.advisor 何靜嫺 zh_TW dc.contributor.advisor Ho, Shirley J en_US dc.contributor.author (Authors) 陳維勛 zh_TW dc.contributor.author (Authors) Chen, Wei-Xun en_US dc.creator (作者) 陳維勛 zh_TW dc.creator (作者) Chen, Wei-Xun en_US dc.date (日期) 2023 en_US dc.date.accessioned 2-Aug-2023 13:42:56 (UTC+8) - dc.date.available 2-Aug-2023 13:42:56 (UTC+8) - dc.date.issued (上傳時間) 2-Aug-2023 13:42:56 (UTC+8) - dc.identifier (Other Identifiers) G0110258025 en_US dc.identifier.uri (URI) http://nccur.lib.nccu.edu.tw/handle/140.119/146478 - dc.description (描述) 碩士 zh_TW dc.description (描述) 國立政治大學 zh_TW dc.description (描述) 經濟學系 zh_TW dc.description (描述) 110258025 zh_TW dc.description.abstract (摘要) 台灣半導體產業在世界上佔重要的地位。在台灣半導體供應鏈上游、中游和下游廠商包含生產超過27種產品。原本傳統的計算方式像是market share或是HHI無法完全的描述整個半導體供應鏈。所以我們用廠商營收的相關性建立三種不同的網路(threshold, MST and PMFG networks)，在檢查這三個網路是否具有實證網路的特徵。我們的研究表明PMFG networks可以最好的描述網路因為網路degree服從power law distribution。使用建構出的PMFG網路計算描述網路和廠商的統計量，包含中心性、加權中心性和cliques。並且利用clique分析進一步將節點依照“production-related” 或 ‘product-related”進行分類，接著利用中心性和專利數量的關係，看是否創新行為是否具有傳染效果。我們的研究發現(1) "production-related" cliques，當cliques至少包含上、中、下游其中一家廠商他的營運成本會比較低。 (2)比較 “cliques producing multiple products” 和 “cliques producing few products” 發現生產比較多產品的cliques他的EPS會比生產比較少產品的cliques來得高。(3)將cliques與金融上的表現連結發現當“three-categories cliques” 裡的廠商股價的平均相關係數最高。 (4)我們研究中心性和廠商專利數量之間的關係，betweenness centrality (BC), closeness centrality (CC), and eigenvector centrality (EC)對於專利數量有正且顯著的影響。 (5) 計算加權的中心性研究加權中心性與專利數量的關係。發現每個節點都使用他的總資產進行加權weighted closeness centrality對於專利數量的影響最大。換句話來說如果每個節點都使用廠商大小來加權我們可以進一步找出影響創新的網路特徵。 zh_TW dc.description.abstract (摘要) Taiwan’s semiconductor industry plays a crucial role in the globe. Our firms cover all upstream, midstream and downstream of the manufacturing process, and provide more than 27 products. Traditional metrics, such as market share or HHI, cannot fully describe the structure of semiconductor industry. As there is lack of actual transactions database, we employ methods that use firms’ revenue correlations to construct three networks (threshold, MST and PMFG networks), then we examine whether the three networks conform to the empirical characteristics of networks. Our analyses suggest that the PMFG network can best describe our data and follow the power law distribution.We then calculate several metrics to measure the network properties and firm-specific characteristics in the constructed PMFG network, including particularly the degrees of centrality and the node-weighted degrees of centrality, and “cliques”. To further investigate the properties of cliques, we classify cliques according to whether the nodes are “production-related” or they are ‘product-related”. Finally, we investigate the relationship between firms’ degrees of centrality and the number of patents they hold, to see if there are contagious effects on firms` innovation activities.Our results show that (1) for "production-related" cliques, if a clique contains at least one firm each from the upstream, downstream, and midstream, then the operating costs tend to be lower. (2) We compare “cliques producing multiple products” to “cliques producing few products”, and found that cliques producing multiple product categories tend to have higher average EPS than those producing fewer categories. (3) We examined the financial connection within cliques, and found that the “three-categories cliques” have the highest average correlation coefficient in member firms’ stock prices. (4) We explored the relationship between the centrality measures of firms and the number of patents, and found that betweenness centrality (BC), closeness centrality (CC), and eigenvector centrality (EC) had a positive and significant effect on patent counts. (5) We calculated the asset-weighted centrality to examine its impact on the number of patents. We found that after each node is weighted by the proportion of total asset, the weighted closeness centrality becomes most relevant for innovations. In other words, after weighing each node by its relative firm size, we can further identify the most relative network feature for innovations. en_US dc.description.tableofcontents List of Tables 6List of Figures 71. Introduction 82 Related Literature 133. Constructing the Networks 183.1 Weights on Edges 203.2 The Threshold Network 213.3 The MST（Minimum Spanning Tree）Network 233.4. The PMFG（Planar Maximally Filtered Graph）Network 254. Verifying the Degree Distributions of Networks 274.1 Scale-free Network and Power Law 274.2 The Threshold Network 294.3 The MST Network 314.4 The PMFG Network 335. Network-Wise Properties and Firm-Specific Characteristics 365.1 Network Properties 365.1.1 Average degree Ek 365.1.2 Network diameter 375.1.3 Network density D 375.1.4 Network centralization C 385.1.5 Network heterogeneity H 385.1.6 Average clustering coefficient E(C) 395.1.7 Power-law exponent r 395.1.8 Assortativity ρ 405.1.9 Modularity Q 415.1.10 Percolation threshold for random node removal fc 425.2 Firm-Specific Network Characteristics 445.2.1 Degree centrality DCi 445.2.2 Closeness centrality (CC(i)) 445.2.3 Betweenness centrality BC(i) 455.2.4 Eigenvector centrality ECi 465.2.5 Node-weighted degree centrality WDCi 475.2.6 Node-weighted closeness centrality WCCi 485.2.7 Node-weighted betweenness centrality WBCi 485.3 Cliques 495.3.1 Production-related Cliques 505.3.2 Product-related Cliques 565.3.3 Cliques and Stock Prices 596. Empirical Analysis: Degrees of Centrality and Patent Counts 606.1 Data and Variables 626.1.1 Number of patents 626.1.2 Other Control Variables 626.2 Regression Results 646.2.1 Contagious effects 646.2.2 Node-Weighted Centrality 667. Conclusion And Suggestion 69References 70Appendix 72 zh_TW dc.format.extent 5146972 bytes - dc.format.mimetype application/pdf - dc.source.uri (資料來源) http://thesis.lib.nccu.edu.tw/record/#G0110258025 en_US dc.subject (關鍵詞) 網絡分析 zh_TW dc.subject (關鍵詞) 冪次分配 zh_TW dc.subject (關鍵詞) 中心性 zh_TW dc.subject (關鍵詞) 集團 zh_TW dc.subject (關鍵詞) 專利數量 zh_TW dc.subject (關鍵詞) Networks en_US dc.subject (關鍵詞) Power Law en_US dc.subject (關鍵詞) Degrees of Centrality en_US dc.subject (關鍵詞) Cliques en_US dc.subject (關鍵詞) Patent Numbers en_US dc.title (題名) 台灣半導體供應鏈：網絡分析 zh_TW dc.title (題名) Taiwan Semiconductor Supply Chain: A Network Analysis en_US dc.type (資料類型) thesis en_US dc.relation.reference (參考文獻) Alstott, J., Bullmore, E., & Plenz, D. (2014). powerlaw: a Python package for analysis of heavy-tailed distributions. PloS one, 9(1), e85777.Arora, A., Ceccagnoli, M., & Cohen, W. M. (2008). R&D and the patent premium. International journal of industrial organization, 26(5), 1153-1179.Audretsch, D. B., & Acs, Z. J. (1991). Innovation and size at the firm level. Southern Economic Journal, 739-744.Birch, J., Pantelous, A. A., & Soramäki, K. (2016). Analysis of correlation based networks representing DAX 30 stock price returns. Computational Economics, 47(4), 501-525.Chen, X., Shangguan, W., Liu, Y., & Wang, S. (2021). Can network structure predict cross-sectional stock returns? Evidence from co-attention networks in China. Finance Research Letters, 38, 101422.Chi, K. T., Liu, J., & Lau, F. C. (2010). A network perspective of the stock market. Journal of Empirical Finance, 17(4), 659-667.Eryiğit, M., & Eryiğit, R. (2009). Network structure of cross-correlations among the world market indices. Physica A: Statistical Mechanics and its Applications, 388(17), 3551-3562.Falasca, M., Zobel, C. W., & Cook, D. (2008, May). A decision support framework to assess supply chain resilience.Hearnshaw, E. J., & Wilson, M. M. (2013). A complex network approach to supply chain network theory. International Journal of Operations & Production Management.Hong, M. Y., & Yoon, J. W. (2022). The impact of COVID-19 on cryptocurrency markets: A network analysis based on mutual information. Plos one, 17(2), e0259869.Huang, W. Q., Zhuang, X. T., & Yao, S. (2009). A network analysis of the Chinese stock market. Physica A: Statistical Mechanics and its Applications, 388(14), 2956-2964.Kenett, D. Y., Tumminello, M., Madi, A., Gur-Gershgoren, G., Mantegna, R. N., & Ben-Jacob, E. (2010). Dominating clasp of the financial sector revealed by partial correlation analysis of the stock market. PloS one, 5(12), e15032.Kim, J., Lee, S. J., & Marschke, G. (2009). Relation of firm size to R&D productivity.Li, Y., Jiang, X. F., Tian, Y., Li, S. P., & Zheng, B. (2019). Portfolio optimization based on network topology. Physica A: Statistical Mechanics and its Applications, 515, 671-681Nerkar, A., & Roberts, P. W. (2004). Technological and product‐market experience and the success of new product introductions in the pharmaceutical industry. Strategic Management Journal, 25(8‐9), 779-799.Nobi, A., Maeng, S. E., Ha, G. G., & Lee, J. W. (2013). Network topologies of financial market during the global financial crisis. arXiv preprint arXiv:1307.6974.Onnela, J. P., Kaski, K., & Kertész, J. (2004). Clustering and information in correlation based financial networks. The European Physical Journal B, 38(2), 353-362.Pakes, A., & Griliches, Z. (1980). Patents and R&D at the firm level: A first report. Economics letters, 5(4), 377-381.POSFAI, M., & BARABASI, A. L. (2016). Network science. Cambridge University Press.S. Wasserman, K. Faust, Social Network Analysis(1994).Cambridge University Press, CambridgeSingh, A., Singh, R. R., & Iyengar, S. R. S. (2020). Node-weighted centrality: a new way of centrality hybridization. Computational Social Networks, 7(1), 1-33.Tumminello, M., Aste, T., Di Matteo, T., & Mantegna, R. N. (2005). A tool for filtering information in complex systems. Proceedings of the National Academy of Sciences, 102(30), 10421-10426.Tumminello, M., Di Matteo, T., Aste, T., & Mantegna, R. N. (2007). Correlation based networks of equity returns sampled at different time horizons. The European Physical Journal B, 55(2), 209-217.White, E. P., Enquist, B. J., & Green, J. L. (2008). On estimating the exponent of power‐law frequency distributions. Ecology, 89(4), 905-912.Wu, L. (2015). Centrality of the supply chain network. Available at SSRN 2651786.Zhang, Y. (2009). Stock market network topology analysis based on a minimum spanning tree approach (Doctoral dissertation, Bowling Green State University).Zhao, L., Wang, G. J., Wang, M., Bao, W., Li, W., & Stanley, H. E. (2018). Stock market as temporal network. Physica A: Statistical Mechanics and its Applications, 506, 1104-1112. zh_TW
