产业集群辨识方法综述

以产业集群内涵的厘清为基础,系统论述并评价了产业集群辨识中所运用到的各种方法,认为:专家意见法、产业感知法和企业访谈法都具有针对性强、能够收集难以统计的时新信息等特点,但依赖于专家等个体感性认知,系统数据收集困难,结论普适化受限;多元统计聚类法透视出产业间重要的依存关系,但部门全覆盖及产业在集群间截然分组不符实际;主成分分析法辨识结果相对理想,但数理解释牵强;Czamanski法逻辑严密,突出集群内部产业间的相互关联,但对支撑性部门雷同的集群处理待改善;共识集群法所体现出来的综合集成理念值得借鉴。最后总结,集群辨识宜继续开发共识集群法所提供的多种分析整合的做法,量化研究为主,定性修正为辅,同时应增加对空间维度的考虑。

REVIEW ON THE METHODOLOGY OF IDENTIFYING INDUSTRIAL CLUSTER

While the significance of industrial clusters for economic development has been widely accepted,there is wide variation in the approaches utilized to identify industrial clusters.In this paper,based on the meaning of industrial cluster,the methods of identifying industrial cluster are systematically discussed and assessed.The relatively qualitative methods,which are expert opinion, industry perception method and survey,are all characteristic of strong pertinency and capability of collecting local current information that is not included in census data,but they rely on experts' or officials' personal knowledge and subjective judgments so heavily that they lack strict rules to make decisions and it's difficult to generalize the results.In the relatively quantitative means, multivariate statistical clustering have insights into the interdependence between industries,however,its results cover the entire industry sectors and show mutually exclusive groups of industries,which don't correspond to reality;principal component analysis' results appear to be satisfying,but according to its algorithm,it tends to produce groupings of industries that have similar input and sales profiles,rather than clusters of industries with strong internal linkage;Czamanski's method could give prominence to the interrelations between industries in a certain cluster,yet it's deficient during identifying the clusters with some same supporting sectors;and consensus clustering's idea of metasynthesis is highly thought to be a breakthrough in methodology.Therefore,we put forward that,in the future research,identifying industrial clusters would follow the modus operandi of metasynthesis in consensus clustering,using quantitative analyses assisted with qualitative approaches,and should place more emphasis on the spatial dimension.