台灣生醫產業研發合作之影響因子與決策規則分析

Analyzing the Determinants and the Decision Rules of R&D Collaboration in the Taiwanese Biomedical Sector

前人研究指出，新興科技產業中，企業仰賴研發合作提升技術研發能力，這現象在生醫產業尤其顯著，因此是否採用合作研發，是生技公司非常重要的決策。然而，目前仍未有足夠文獻探討這些公司如何整合衡量自身的特性，以擬定與外部機構進行合作研發之決策，包括地理位置、技術類型、資源與過去經驗。本研究採用蟻群尋優法能夠篩選重要因子、萃取決策規則的優勢，針對我國新興生醫產業112間公司於1998-2014年間之實證資料，以宏觀角度探究影響公司採取研發合作決策的特性因子，及因子組成的決策規則。此外，本研究亦以2007年生技新藥產業發展條例之頒布為例，比較條例施行前後之差異，並針對公司進行產學合作之決策，共萃取出四組規則，正率皆高於90%。四組規則皆顯示過去合作經驗為主要的因子，而與技術類型、研發能力、政府資金投入等因子所組合的規則皆會影響公司合作研發、產學合作。本研究採蟻群尋優法萃取分類規則，找出生醫公司進行合作決策的可能整合考量，並分別訪談生技、醫材、製藥三大產業的專家驗證這些規則在業界的意義與適用性，以提供未來研究者與政策規劃者在相關議題之探究方向，以較有效的促進產業創新。

Literature maintains, in the high-tech emerging sectors, firms heavily rely on R&D collaboration to enhance their R&D capabilities. The innovation network literature has indicated that innovation is embedded in the networks instead of any single actor (such as a firm), especially in the case of the biomedical sector. In the high-tech sectors, the determinants of firms to establish R&D collaborations with other firms or institutes by considering the characteristics such as geographical proximity, technology domain, potential resources and the track records. However, what less clear are the determinants of building collaboration networks in the emerging sector. Combining Ant Colony Optimization (ACO) and interviews, this study explores the decision rules of R&D collaborations in the biomedical industry, a fast growing high-tech sector. The ACO algorithm is a probabilistic technique for finding suitable paths and extracting the decision rule. This study, therefore, applied ACO algorithm to macroscopically extract important attributes and decision rules composed of firm features from 112 biomedical firms. This study also compared the rule sets from the analyses of the two periods before and after the implementation of Biotech and New Pharmaceutical Development Act in 2007. Furthermore, the decision rules of academia-industry R&D collaborations in the biomedical sector are also analyzed. The 4 rule sets with accuracy rate higher than 90% show that prior collaboration experience is the most important attribute. For validating the meaning of the rules in reality, we conducted interviews with experts in the field of biopharmaceuticals, medical-device, and pharmaceuticals. The mutual complementation of technology, R&D capabilities, and government funding supports also influences the determinants of R&D collaborations. Finally, this study offered policy suggestions to effectively facilitate innovation in the emerging sectors.