초록 ▼

1. 서론 : 연구의 배경 및 목적
우리나라의 지역산업정책은 클러스터 기반의 정책을 실시해 왔다. 혁신 인프라 구축, 연구개발, 인력양성 등 여러 지원프로그램을 하나로 묶은 패키지형태의 사업을 추진해왔다. 그리고 지원사업의 경우 전 주기적으로 지원하였다. 그리고 전략산업을 선정하고 이에 자원을 집중하는 선별적 산업정책을 실시해왔다. 이러한 지역산업정책의 추진 과정에서 다음과 같은 문제점이 제기된다.
첫째, 클러스터 기반의 정책과 선별적 산업정책은 특정 클러스터와 특정 산업을 육성 대상으로 삼는다는 점에서는 공통점이 있으나

1. 서론 : 연구의 배경 및 목적
우리나라의 지역산업정책은 클러스터 기반의 정책을 실시해 왔다. 혁신 인프라 구축, 연구개발, 인력양성 등 여러 지원프로그램을 하나로 묶은 패키지형태의 사업을 추진해왔다. 그리고 지원사업의 경우 전 주기적으로 지원하였다. 그리고 전략산업을 선정하고 이에 자원을 집중하는 선별적 산업정책을 실시해왔다. 이러한 지역산업정책의 추진 과정에서 다음과 같은 문제점이 제기된다.
첫째, 클러스터 기반의 정책과 선별적 산업정책은 특정 클러스터와 특정 산업을 육성 대상으로 삼는다는 점에서는 공통점이 있으나 선별적 산업정책은 특정 산업을 보호하고 육성하는데 초점을 두지만 클러스터 기반 정책은 관련 산업의 연계와 가치사슬 형성을 강조한다. 산업정책은 좁은 범위의 산업에 초점을 두지만, 클러스터 기반 정책은 전후방 연관 산업의 가치사슬 형성과 연계를 강조하여 지원 범위가 더 넓다고 할 수 있다. 따라서 선별적 산업정책에서 클러스터 기반 정책으로 전환할 필요가 있다.
둘째, 클러스터 기반의 정책을 수행하면서 클러스터의 식별 및 분석이 필요한데, 지금까지는 산업 수준에서 특화도(입지계수), 집중도(전국대비 비중), 성장성, 생산성 등을 이용하여 분석하였다. 그러나 연관 산업의 지리적 집중이라는 클러스터의 식별을 위해서는 산업 간 연관성을 측정하는 지표를 명시적으로 포함시킬 필요가 있다.
셋째, 산업 간 연계를 검토한 경우에도 특정 지역을 대상으로 네트워크 등을 분석하였다. 특정 지역을 대상으로 할 경우 해당 지역의 산업 간 연계 등을 볼 수 있는 장점이 있으나 다른 지역과 비교할수 없는 단점이 있다. 이런 점에서 여러 지역을 비교할 수 있는 클러스터 식별 방법이 필요하다.
이런 문제점을 극복하고자 본 연구는 첫째, 연관 산업의 지리적 집중이라는 관점에서 클러스터를 식별하고 분석한다. 이를 위해 산업 간 연계성을 고려하는 지표로 투입산출연계, 직업의 유사성, 사업체 및 고용의 입지상관관계 등을 명시적으로 포함한다. 둘째, 그렇게 식별된 클러스터가 어느 지역에 분포되어 있는지를 매핑을 통해 분석한다. 셋째, 클러스터 식별 방법에서 특정 지역 수준의 클러스터가 아니라 전국의 여러 지역을 비교할 수 있는 클러스터를 식별하려고 하였다. 그렇게 함으로써 여러 지역을 상호 비교할 수 있다.넷째, 클러스터가 지역경제에 미치는 영향이 어떤지를 분석한다.클러스터는 외부성에 의해 긍정적 효과도 있으나 혼잡비용 등과 같은 외부 불경제에 의해 부정적 효과도 있다. 이에 대해 분석하여 클러스터 기반의 지역산업정책에 미치는 시사점을 제시한다.

Abstract ▼

□ Research background and objectives
Korea’s regional industrial policies have been implemented based on clusters. Projects have taken place in the form several support programs including establishment of innovative infrastructure, research and development, manpower training, etc. In addition, su

□ Research background and objectives
Korea’s regional industrial policies have been implemented based on clusters. Projects have taken place in the form several support programs including establishment of innovative infrastructure, research and development, manpower training, etc. In addition, support projects have received assistance throughout all periods. Moreover, selective industrial policies targeting strategic industry and resources have been implemented. In the implementation process of these regional industrial policies, the following problems have been brought up.
First, although cluster-based policies and selective industrial policies have similarities in that they target a specific cluster and specific industry for promotion, selective industrial policies focus on protecting and fostering a specific industry while cluster-based policies focus on linking related industries and creating a value chain. Therefore, there is a need to shift from selective industrial policies to cluster-based policies.
Second, when implementing cluster based policies, the identification and analysis of clusters is necessary. This was done by analyzing the specialization degree (measured by location quotient), concentration degree (the share of national employment), rate of growth, and productivity on the industry level. However, in order to identify clusters in aspects of geographical concentration of related economic activities, it is necessary to explicitly consider the inter-industry linkages.
Third, network analysis of specific regions was conducted even when examining the linkage among industries. In targeting a specific region, there are advantages when identifying the linkage among industries of the region, but there are also difficulties when comparing with other regions. Therefore, a cluster identification method that enables comparison with other regions is necessary.
This research has been conducted in consideration of the following purposes. First, clusters should be identified and analyzed from the viewpoint of the geographical concentration of related industries. To do this, indicators that consider the linkage among industries such as the input-output linkage, job similarities, location correlations of business and employment should be explicitly included. Second, the locational distribution of the identified clusters should be analyzed by the mapping method. Third, instead of identifying clusters in specific regions, clusters are identified on the national level so that comparison among regions can be possible. Fourth, the performance of clusters in the regional economy was analyzed. Although clusters may have positive effects through externality, it also has negative effects of external diseconomy such as congestion costs. By analyzing the cluster effects, the implications of cluster-based regional industrial polices are presented.
□ Cluster mapping analysis
In order to identify clusters, we have conducted the following process referring to Delgado et al.(2014).
First, industries were newly classified by linking data between 5-digit KSIC (Korean Standard Industrial Classification) industries, 3-digit KSCO(Korean Standard Classification of Occupations) occupations, and input- output tables. As a result, a total of 327 industries were identified.
Second, by referring to Porter(2003), industries were divided into traded industries, local industries and resource dependent industries, and clusters were identified concentrating on traded industries. Industries were classified into traded industries using the specialization and concentration patterns; the share of employment in regions with LQ ≧ 1 is greater than 50% of total employment of the industry; the top 10% regions in terms of LQ hold more than 25% of total employment of the industry. Traded industries numbered a total of 211.
Third, the similarity matrix was made by using 4 indicators including the input-output linkage, job similarity, location correlation of establishment and employment.
Fourth, preliminary clusters were identified by applying the similarity matrix and the initial number of clusters differently several times.The validity score of the preliminary cluster was used to finally identify the clusters. The cluster validity score and industrial validity score were used. There turned out to be 30 preliminary clusters.
Fifth, clusters that have been classified quantitatively were judged qualitatively with adjusted outliers. Finally, 26 clusters have been identified.
To examine the spatial distribution of the 26 clusters, mapping was used for visualization. This visualization increases the understanding of the spatial distribution of clusters and helps spatially verify the validity of identified clusters. The locational quotient of over 1.25 and the national proport exceeding a certain level, such as 1% and 3%, were used as criteria. The mapping result of the 26 clusters have been marked on a map.
□ Cluster impact analysis
The economic outcome of the identified clusters influencing the regional economy was analyzed. The cluster performance analysis is as follows. The employment growth rate of industry i in a specific region may be explained by the degree of early-stage development, the degree of cluster development, and the degree of neighboring regions' cluster development of industry i. Dummy variables of region r and industry i are included.
The result of cluster effects analysis shows that the employment growth rate of the regional industry decreased as the level of agglomeration of the relevant industry increased (-0.250). On the other hand,the employment growth rate of the regional industry increased as the agglomeration of the regional clusters increased (0.137), and also as the clusters of neighboring regions were strengthened (0.112). In other words, it turned out that the stronger the cluster, the more the regional economy grew,in addition to greater promotion of the neighboring region's economy.
It was also examined whether clusters effect the increase in the number of businesses of the regional industry. The estimation result was that clusters have a significant effect on the increase of businesses of the regional industry (0.049), as well as neighboring clusters (0.040).
As shown above, there is a positive effect on the relevant region and neighboring regions’ economic growth corresponding to the strength of the regional cluster.On the other hand,thers is conditional convergence in early stage development of the regional industry,which has a negative effect on the regional economy.
Thece results are qualitatively similar on the province level.Furthermore,the effect of clusters of the regional industry was bigger in regions with larger population.Cluster effects were bigger in regions with higher levels of technology and also regions that had core clusters.These results provide various implications for setting regional industrial policy direction.
□ Policy implications
First, there is a need to transfer regional industrial policies from the industrial approach to a cluster approach. Until now, a selective industrial policy of selecting a strategic industry and investing in it has been in effect. In order to implement cluster policies, a wider scope of regional industrial polices that include related industries and take into account the linkage among industries is necessary. One of the key points of this research is that the diffusion effect among related economic activities is the core engine of growth and employment. Therefore, it is advisable to switch from a narrow industrial approach to a cluster approach that emphasizes linkage with related industries.
Second, it is necessary to explicitly consider the industrial and spatial correlation in cluster identification. In cluster identification, the linkage between industry and space should be explicitly considered, rather than the specialization degree or concentration degree of industry level. In this research, the similarity matrix has been made by using the location correlation coefficient of businesses and employment, job similarity, and industrial linkages. This method better corresponds with the definition of clusters as a geographical concentration of related industries.
Third, cluster-based policies are justified. The positive effects of clusters have been verified, and therefore cluster policies for regional industrial policies have been proven as effective tools. The stronger the cluster, the more positive effects on the region and neighboring regions’ economic growth have been shown. Therefore, the regions’ assets, potential, and the strengths of existing clusters should be utilized as leverage and related diversification should be pursued.
Fourth, it is necessary to form a cooperative relationship with neighboring regions in cluster policy. Clusters have positive effects on not only the relevant region, but also neighboring regions. The scope of positive effects appeared not only in the city and town units, but also in the province units. Therefore, rather than a competitive relationship with neighboring regions, a complementary and cooperative relationship is advised.
Fifth, diversification and specialization should be promoted at the same time. If clusters can be translated to a geographic concentration of related economic activity, cluster development may be translated to related diversification. A more preferable cluster policy is pursuing diversification and specialization simultaneously, which can go beyond the dichotomy of localization and urbanization as the source of externality.When approaching industrial policies with the cluster concept, related diversification is possible and industrial policy could overcome the dichotomy of selective policy and horizontal policy. In pursuing related diversification, cluster policies and smart specialization strategies share a common objective.