2023 Volume 11 Issue 2 Pages 126-149
Empirical studies on agro-industrial complexes are few. Facing the issue of their sustainability, the vitality of the complex is analyzed in South Korean settings through structural equation modeling of the survey data of 600 companies. Unlike the existing literature on regional economics, the survey evaluates objective characteristics at the local, complex, and company levels while incorporating psychometrics, subjective qualities of the local environment, the complex itself, infrastructure, networking, and supportive policy. Among the objective characteristics, significant variables in the employment and sales models turn out to be consistent and mainly at the company level. Among the subjective variables, employment size and sales are associated with the qualities of infrastructure and supportive policy. Specifically, potential employees will be attracted by accessible transportation systems to neighboring major cities rather than convenience facilities within the complex. Sales may be improved by providing direct support for the marketing of each company compared to indirect support regarding agro-industrial complexes, infrastructure, and networking.
In the 1960s, the Korean military regime began to develop large-scale industrial complexes to restructure and transition its industry from simple farming and fishing to high-value-added manufacturing. Firstly, on January 20, 1962, the government enacted the “Act on Special Cases Concerning Land Expropriation for the Development of Industrial Complexes,” enabling industrial complexes to be appointed by a cabinet order. As such, seven days later, the government selected Ulsan City to develop the first such complex; it is currently known as the Ulsan Mipo National Industrial Complex (Cabinet Order 403), in which heavy chemical (Hyundai Heavy Industries), petrochemical (SK Energy), and automobile (Hyundai Motor Company) industries were intensively promoted. As of June 17, 2020, Korea has 1,221 industrial complexes (47 national industrials, 672 general industrial, 30 urban high-tech industrial, and 472 agro-industrial complexes). According to industry types, light industries were located in the Korean Capital Region because of affluent human resources and heavy industries in southeast coastal cities (e.g., Ulsan, Busan, and Pohang) to import raw materials.
Meanwhile, the agro-industrial complex has unique characteristics, unlike the other three industry types above. It is defined as the “industrial complex designated … to attract and promote industries for increasing the incomes of farmers and fishermen in rural areas” [Article 2 of the “Industrial Sites and Development Act” (ISDA)]. Specifically, its development was officially promoted in the 1980s to utilize resources of idle human labor during the off-season, increasing incomes by advancing industrial structures in rural areas. In the 1960s and 1970s, the urban-rural divide was aggravated by the national preferential policy for the manufacturing industry. In the late 1970s, the growth of agricultural incomes was substantially reduced because of agricultural trade liberalization. Thus, instead of the “Saemaul Factory Building Project” in the 1960s and 1970s, as part of the “Saemaul (New Village) Movement,” a more comprehensive support project for agro-industrial complex development was initiated (Lee, D.-P. and Lee, 1996). In 1981, the government launched the Development Planning Office for Non-Agricultural Income Sources, and the “Act on the Promotion of Income Source Development for Agricultural and Fishing Villages” was enacted in 1983. In 1984, seven complexes were firstly established throughout the country. As of June 17, 2020, 7,511 companies are in 472 agro-industrial complexes. They have contributed to mitigating unbalanced, Seoul-centered development by redistributing industries and vitalizing the regional economy in rural areas (Gim, Choi et al., 2020).
Figure 1 shows a choropleth of the number of agro-industrial complexes at the unit of the district (si, gun, and gu) whose head is in charge of designating the complexes (Along with the choropleth, point maps (PNG) of the locations of the complexes (and companies stored in a KICOX database) and as their bases, enlargeable Google maps (HTML)—thus, the exact locations can be identified—as well as kernel density maps (PNG) are available online for downloading: https://drive.google.com/file/d/1ZRX1LckA8_5xVurTqMsTQxx1Qkx6d6ak/view?usp=sharing. ).
Currently, however, along with its rapidly decreasing birth rate and steady economic growth, Korea faces a threat to the sustainability of its agro-industrial complexes that have been developed since the economic transition era of the 1980s. For about four decades after the first establishment in 1984, however, studies on these agro-industrial complexes were extremely few, while no studies for international readership exist. Specifically, a total of three domestic articles from the literature—Lee, C.-W. (2008); Lim, An et al. (2010); Woo, J.M. (2008)—presented issues with Korean agro-industrial complexes (Gim, Choi et al., 2020; Lee, K.-R., 2018). Only six professional reports by two national research institutions on rural economy and regional planning reviewed the issues and vitality of the complexes:Bae and Kim (2005); Choi, Kim et al. (2014); Choi, K., Kim, Y. et al. (2012); Jang, C.-S. (2011); Lee, D.-P. and Lee (1996). Kim, S.-B. and Hong (2010). Until 2020, when a consulting project (Gim, Choi et al., 2020) was performed, even the Korea Industrial Complex Corporation (KICOX), as the national coordinator for industrial complexes, had not yet attempted a comprehensive review of the challenges of the agro-industrial complexes (Lee, K.-R., 2015). Nevertheless, in 2008, it published “A Handbook on the Development and Support Services for Agro-Industrial Complexes.”
In an effort to strengthen agro-industrial complexes' functional role as a regional economy's driving force, further studies and political approaches are required to examine factors affecting spatial revitalization and economic growth.
At this juncture, this study intends to be the first systematic academic research on Korean agro-industrial complexes. A difference from the few Korean domestic studies is that this study uses primary survey data through probability sampling to explore what affects the complex vitality—evaluated with employment and sales—instead of secondary data. For this purpose, its empirical analysis employs an inferential technique. As a step further, the empirical model measures subjective variables through psychometrics on the qualities of the complex’s neighboring environment, the complex itself, its infrastructure, networking, and policy, as well as objective variables that have been consistently dealt with in studies on usual industrial complexes (Huh and Lim, 2012; Sergi, Popkova et al., 2019; You and Byun, 2011).
Korean agro-industrial complexes—and rural areas overall (Yi and Son, 2022)—currently face sustainability challenges according to rapidly changing socioeconomic settings. Choi, Kyeonghwan, Kim, Yonglyoul et al. (2012) noted their inherent weakness from the perspective of industrial location theories since, in principle, they were forced to be built in far-flung rural areas rather than those with large populations (for markets and human resources), affluent raw materials, or convenient transportation systems.
The second issue is population aging in rural areas. The nationwide decline in the birth rate—the total fertility rate in Korea is 0.75 as of the second quarter of 2022, which is among the lowest in the world—and rural-to-urban migration of the younger population aggravates the difficulty of complex companies hiring employees (Chen and Akita, 2021). Because of the population decline, 39.7% of the neighborhoods, the lowest administrative unit, and 84.1% of the total 82 districts—the district is a group of neighborhoods—are expected to disappear by the mid-2040s (Park and Kim, 2016). The workforce shortage in agro-industrial complexes due to the population decline and migration is also common among post-Soviet republics (e.g., Russia, Kazakhstan, Ukraine, and Belarus); in fact, most studies on the complexes have been done in these countries (NEDELKIN, NOVIKOV et al., 2017).
Third, historical emphasis was placed on launching new agro-industrial complexes rather than facilitating their operations. By quickly locating unrelated industries due to pressure on lot sales, synergies between complex companies and local industries hardly occurred (Choi, Kim et al., 2014). This is partially due to provisions in Article 8 of the ISDA, which outlines the appointment of the agro-industrial complex by district mayors. Other types of complexes can be appointed by province governors and/or the Minister of Land, Infrastructure, and Transport. In contrast to central and provincial governments, local governments tend to appoint agro-industrial complexes for political purposes instead of systematically establishing long-term plans for flexible responses to industrial restructuring (Jang, I.-S., Kim et al., 2012).
Concerning their ages (Kim, K., 2005) and scales (Lee, K.-R., 2015), the most noted issue with Korean agro-industrial complexes is outdated and lacking infrastructure and facilities. As stated above, the undue emphasis on opening new agro-industrial complexes made undervalue greenspaces, roads, and employee-supportive functions (Huh and Lim, 2012). As of the fourth quarter of 2019, 62% of the complexes are 20 years or older (292 out of the total 472 complexes), and 76% of the total companies are located in these old complexes. Also, among all types of industrial complexes, agro-industrial complexes occupy 39% in number but just 5% in size: As of June 17, 2020, their mean area is 163,000 m2 (compared to 17,159,000 m2 for national industrial complexes, 799,000 m2 for general industrial complexes, and 279,000 m2 for urban high-tech industrial complexes). As main advantages, such a small-scale development (below 150,000–200,000 m2) is exempt from the requirements of the Environmental Impact Assessment (as in the ISDA), Establishment of the Energy Use Plan, and Transportation Impact Assessment (according to the Act on Special Cases Concerning the Simplification of Authorization and Permission Procedures for Industrial Complexes). However, the small complex is noted by few parking lots, access roads, ill-managed employee facilities, and community parks. The lack of amenities, in turn, negatively works on attracting young employees.
The fifth issue frequently identified in professional reports by government-affiliated research institutions (Bae and Kim, 2005; Choi, K., Kim, Y. et al., 2012; Kim, S.-B. and Hong, 2010) is the shortage and ineffectiveness of government support policies (Gim, Choi et al., 2020). It results primarily from a diversified management system. Multiple acts deal with agro-industrial complexes—specifically, the ISDA, the ICDFEA, the Agricultural and Fishing Villages Improvement Act, and the Integrated Guidelines for Development and Operation of Agricultural Industrial Complexes—so unlike national industrial and general industrial complexes, no coordinating/responsible ministry exists in the central government for systematically supporting agro-industrial complexes: Six ministries and two state-owned enterprises [the Ministry of Trade, Industry and Energy (MTIE), the Ministry of Land, Infrastructure, and Transport (MLIT), the Ministry of Agriculture, Food and Rural Affairs/Ministry of Oceans and Fisheries, the Ministry of Environment, the Ministry of the Interior and Safety, the Korea SMEs and Startups Agency, and the KICOX] co-share the responsibility for agro-industrial complexes. The responsibility is concerned mainly with developing new complexes rather than facilitating their performance and providing customized support for complex companies (Choi, Kim et al., 2014). As stated above, the complexes are designated by district mayors, so they are ineligible for programs to regenerate and advance old industrial complexes. The programs are targeted at those under the jurisdiction of the MLIT and MTIE (Choi, K., Kim, Y. et al., 2012; Lee, K.-R., 2015), and district governments do not have enough budget for their programs (Chang and Lee, 2016).
Last, competitive disadvantages relative to other types of industrial complexes have been noted (Choi, Kim et al., 2014). In line with national growth, highways have been built and expanded in rural areas to the degree to which they can accommodate national/general industrial complexes that are designated and intensively supported by the central/provincial governments. Thus, the agro-industrial complex's lot sales price and image became uncompetitive. This negatively affected securing buyers and employees.
Concerning the above six issues of Korean agro-industrial complexes, a few studies attempted to evaluate the sustainability of Korean agro-industrial complexes. Lee, C.-W. (2008) conducted a case study of a complex in Goryeong-gun through a survey of company representatives and found that the main advantage is low land prices. However, since general industrial complexes became located near the agro-industrial complex (Choi, Kim et al., 2014), he argued that for sustainability, priorities should be placed on facilitating innovations through industry-academia cooperation, strengthening networks between related companies, and building community social capital for general support.
Woo, J.M. (2008) explored policy demands of agro-industrial complexes in the North Chungcheong Province by summarizing 2006 North Chungcheong statistical data and reviewing existing reports, particularly their analytical results (numbers of the complexes/companies, lot sales rate, operation rate, output, and complex area). Then, he highlighted the importance of governmental support for specialized/area-specific businesses, innovations, employment/human resources development (HRD), and complex association activities.
Lim, An et al. (2010) categorized Korean agro-industrial complexes by size and growth rate to analyze their spatial distribution patterns. The patterns were found not to be systematic and attributed to their individual development without consideration of regional characteristics. Accordingly, cooperation was suggested between the complexes in the same region.
The above studies examined agro-industrial complexes with the descriptive statistics of secondary data at the complex level or survey data from a limited number of companies in the same complex/province. In contrast, this study extensively collected primary data across the entire country (for higher external validity) at the regional, complex, and company levels (complex-level secondary data were also used to confirm the reliability). Also, the primary survey data were collected by probability sampling (for sample representativeness) and analyzed with inferential statistics (for higher internal validity). Another difference from previous studies is that this study’s survey items measured objective characteristics and subjective qualities of the local environment, the complex, its infrastructure, networking, and supportive policy; the subjective items were analyzed using a psychometric method.
To evaluate the vitality of agro-industrial complex companies, this study employed two variables, according to You and Byun (2011): employment size and sales. Two possible determinants were comprehensively investigated by reviewing prior studies, including the above-discussed academic papers and professional reports. Regarding variable types, Bae and Kim (2005) categorized issues with agro-industrial complexes into the complex itself (e.g., human resources—such as local population aging—transportation conditions, and accessibility), complex operation (e.g., supportive systems), legal system (e.g., development and management systems), and complex business operations (e.g., ownership). Kim, S.-B. and Hong (2010) identified various issues in relation to complex locations (e.g., transportation and employment), geographical accessibility, complex age and scale, industrial specialization, connection to local resources, variety of public policies, infrastructure, and cooperation between industries/companies/complexes. Choi, K., Kim, Y. et al. (2012) noted the issues of complex scale, access roads, infrastructure, employee support (lack of facilities and commuting modes), public policy, and connection to or accessibility of local resources. This study arranged all these issues and determined research variables through group interviews and written consultations with 24 experts representing 10 universities and 6 research institutions. The conceptual model of this study is shown in Figure 2. This study hypothesized that employment size and sales are affected by different factors. Specifically, we expected increased employment size to be led by the quality of infrastructure and transportation accessibility. Sales would depend mainly on governments’ supportive policies and networking for aggregation.
Table 1 shows predicting variables, including location (four dummies for the total five regions), accessibility (expressways and harbors), and human resources (high schools and colleges) on the regional scale and as complex-scale variables, temporal (designated year), spatial (designated, managed, total, and company-average areas), and activity-level (lot sales and operation rates and average employment, production, and exports) characteristics. On the company scale, this study also evaluated temporal (years of establishment and complex location) and spatial (site and building areas) features to reflect the general business competitiveness; it used dummy variables for company ownership, industry type, and whether the company received policy support and conducted research and development activities (R&D). Industry type was measured with four dummies according to five categories according to the Korean Standard Industrial Classification. Notably, as suggested during the expert interviews and consultations, this study evaluated the influences of subjective qualities on the company vitality using 50 indicators on the local environment (8 formative indicators), the complex itself (7), infrastructure (9), networking (9), and supportive policy (12) (50 = 45 formative indicators + 5 items on the general quality of each of the five dimensions).
| Min | Max | Mean | S.D. | ||
|---|---|---|---|---|---|
| Continuous: objective [units] | Employment size (EmpNu) [persons] | 1 | 643 | 27.4383 | 48.3255 |
| Sales (TSal9) [million KRW] | 6 | 280,000 | 12,037.7246 | 29,239.5142 | |
| Expressway interchanges (Expway) [counts/10km radius] | 0 | 53 | 10.7317 | 9.1611 | |
| Harbors (Harb) [counts/10km radius] | 0 | 2 | 0.1700 | 0.3890 | |
| High schools (Sch_Hi) [counts/province] | 0 | 142 | 11.5817 | 14.6226 | |
| Colleges (Sch_Co) [counts/province] | 0 | 16 | 2.0500 | 2.5895 | |
| Complex designated year (Yr_Cx) [year] | 1,984 | 2,013 | 1,994.4067 | 8.4101 | |
| Complex designated area (Are_As) [1,000 m2] | 34 | 568 | 200.2317 | 105.7267 | |
| Complex managed area (Are_Ma) [1,000 m2] | 2 | 568 | 198.6317 | 106.6731 | |
| Complex total area (Are_To) [1,000 m2] | 29 | 481 | 155.2967 | 85.6969 | |
| Complex company-average area (Are_Av) [1,000 m2] | 0.01 | 53.00 | 6.8250 | 5.2884 | |
| Lot sales rate (LotS) [%] | 33.00 | 100.00 | 97.9304 | 9.2953 | |
| Operation rate (Oper) [%] | 0.01 | 1.00 | 0.8952 | 0.1457 | |
| Average employment size (Emp_Av) [persons] | 0.01 | 139.75 | 19.4328 | 17.6729 | |
| Average production amount (Sal_Av) [1,000 KRW] | 0.00 | 68,155.67 | 2,251.9763 | 5,719.8438 | |
| Average export amount (Exp_Av) [USD] | 0.00 | 20,515.60 | 353.9000 | 1,574.3179 | |
| Company established year (Yr_CyE) [year] | 1,957 | 2,019 | 2,003.6450 | 10.7313 | |
| Company moved year (Yr_CyM) [year] | 1,944 | 2,019 | 2,008.7713 | 8.7226 | |
| Company site area (Are_Si) [m2] | 41 | 1,045,634 | 9,788.3290 | 45,990.6660 | |
| Company building area (Are_Bu) [m2] | 0 | 7,658,312 | 15,984.4100 | 312,599.6259 | |
| Continuous: subjective [1 (strongly disagree)–5 (strongly agree)] | Accessibility is good to the product market. (Set_1) | 1 | 5 | 3.3840 | 0.6912 |
| Purchasing materials and components is easy. (Set_2) | 1 | 5 | 3.5576 | 0.6279 | |
| Accessibility is good to expressways. (Set_3) | 1 | 5 | 3.5192 | 0.6302 | |
| Settings are convenient for exportation (e.g., harbors are close and sea routes are secured). (Set_4) | 1 | 5 | 2.9265 | 0.5559 | |
| Labor hiring is easy. (Set_5) | 1 | 5 | 2.7145 | 0.8036 | |
| Settings are favorable for searching for new markets. (Set_6) | 1 | 5 | 2.8948 | 0.5338 | |
| Networks are built with nearby high schools, colleges, and research institutions. (Set_7) | 1 | 5 | 2.7312 | 0.5894 | |
| Residential settings are established for complex workers to live in. (Set_8) | 1 | 5 | 2.6177 | 0.7170 | |
| Overall, the regional settings of the complex are satisfactory. (Set_G) | 1 | 5 | 3.0690 | 0.4931 | |
| Site area in the complex is sufficient. (Com_1R) | 1 | 5 | 3.4850 | 0.6908 | |
| Agglomeration effects occur because companies in specific industries are concentrated. (Com_2R) | 1 | 4 | 2.7483 | 0.9415 | |
| Information exchange is adequate among complex companies. (Com_3R) | 1 | 5 | 3.0500 | 0.6794 | |
| Landscape in the complex is good. (Com_4R) | 1 | 5 | 2.9617 | 0.7446 | |
| Environmental management around the complex is well done. (Com_5R) | 1 | 5 | 3.0667 | 0.6580 | |
| Complex association activities are active. (Com_6R) | 1 | 5 | 3.0900 | 0.6972 | |
| Complex management is systematically done (e.g., management office and responsible government officials). (Com_7R) | 1 | 5 | 2.9033 | 0.8748 | |
| Overall, the complex settings are satisfactory. (Com_G) | 1 | 5 | 3.1414 | 0.5780 | |
| Waste disposal facilities are well established within the complex. (Inf_1R) | 1 | 5 | 1.8083 | 1.1535 | |
| Electricity supply is adequate within the complex. (Inf_2R) | 1 | 5 | 3.9933 | 0.4586 | |
| Gas supply is adequate within the complex. (Inf_3R) | 1 | 5 | 1.6200 | 1.1566 | |
| Water supply is adequate within the complex. (Inf_4R) | 1 | 5 | 3.9600 | 0.5312 | |
| Logistics facilities are enough within the complex (for storage, transportation, handling, etc.). (Inf_5R) | 1 | 5 | 2.2233 | 1.3428 | |
| Roads are adequate for vehicle travel. (Inf_6R) | 1 | 5 | 3.5100 | 0.7005 | |
| Parking space is sufficient. (Inf_7R) | 1 | 5 | 2.9150 | 1.1060 | |
| Community facilities are established for complex workers. (Inf_8R) | 1 | 5 | 2.0767 | 0.9569 | |
| Residential facilities (dorms) are located for complex workers. (Inf_9R) | 1 | 5 | 1.3650 | 0.8203 | |
| Overall, the infrastructure of the complex is satisfactory. (Inf_G) | 1 | 5 | 3.4501 | 0.6718 | |
| Joint programs with nearby colleges and research institutions are well conducted. (Net_1) | 1 | 5 | 2.6560 | 0.6224 | |
| A cooperative relationship is formed between complex companies. (Net_2) | 1 | 5 | 3.1390 | 0.6096 | |
| Facilities and policies are available for cooperation between complex companies. (Net_3) | 1 | 5 | 2.8576 | 0.6591 | |
| A network is formed with associate companies in/near the region. (Net_4) | 1 | 5 | 2.9531 | 0.6302 | |
| A cooperative relationship is established with the local (si/gun/gu) government. (Net_5) | 1 | 5 | 3.1960 | 0.6879 | |
| A cooperative relationship is established with the provincial (do) government. (Net_6) | 1 | 5 | 3.0988 | 0.7272 | |
| A department/institution operates in the provincial (do) government for the complex management and support. (Net_7) | 1 | 5 | 3.0871 | 0.6823 | |
| A cooperative relationship is established with related ministries (e.g., Ministry of Trade, Industry and Energy, Ministry of Agriculture, Food and Rural Affairs, Ministry of the Interior and Safety, and Ministry of SMEs and Startups). (Net_8) | 1 | 5 | 2.8241 | 0.7054 | |
| A cooperative relationship is formed with local residents. (Net_9) | 1 | 5 | 2.9079 | 0.5645 | |
| Overall, the network settings of the complex are satisfactory. (Net_G) | 1 | 5 | 2.9780 | 0.5411 | |
| Location and investment grants are sufficiently offered. (Sup_1) | 1 | 5 | 2.7933 | 0.5487 | |
| Tax and financial benefits (e.g., tax reduction) are satisfactory. (Sup_2) | 1 | 5 | 2.8933 | 0.5439 | |
| Local (si/gun/gu) government policy for supporting complex companies is satisfactory. (Sup_3) | 1 | 5 | 3.0283 | 0.6012 | |
| Public relations and marketing policy for supporting the complex and companies are well established. (Sup_4) | 1 | 5 | 2.6983 | 0.6256 | |
| Information on policies and benefits for complex companies is continuously offered by the provincial (do) government or other administrative agencies. (Sup_5) | 1 | 5 | 2.9200 | 0.7196 | |
| Technology support programs are operated for complex companies, such as new technology training sessions. (Sup_6) | 1 | 5 | 2.5617 | 0.5803 | |
| Supportive agencies (or policies) exist for establishing the cooperative relationship between complex companies and external research institutions/colleges. (Sup_7) | 1 | 5 | 2.6400 | 0.6146 | |
| Local government support is provided in hiring employees (e.g., ads, public relations, and related policies). (Sup_8) | 1 | 5 | 2.8100 | 0.6792 | |
| Administrative procedure for moving into the complex is easy. (Sup_9) | 1 | 5 | 3.0917 | 0.5720 | |
| After the move, information for the settlement was sufficiently offered. (Sup_10) | 1 | 5 | 3.0533 | 0.6938 | |
| Local government is active in building community facilities for complex workers (e.g., operating commuter buses). (Sup_11) | 1 | 5 | 2.4067 | 0.6623 | |
| Problems are easily addressed because of the rapid response by the government (e.g., public servi with complaints). (Sup_12) | 1 | 5 | 2.6283 | 0.6590 | |
| Overall, settings to support the complex are satisfactory. (Sup_G) | 1 | 5 | 2.8502 | 0.5286 | |
| f | % | ||||
| Categorical: objective | Location (LocR) | Gangwon (1) | 101 | 16.83 | |
| Chungcheong (2) | 135 | 22.50 | |||
| Southwest (3) | 179 | 29.83 | |||
| Daegu–North Gyeongsang (4) | 91 | 15.17 | |||
| Southeast (5) | 94 | 15.67 | |||
| Company ownership (Own_Y) | No (0) | 128 | 21.33 | ||
| Yes (1) | 472 | 78.67 | |||
| Industry class (InduR)† | 1 | 90 | 15.00 | ||
| 2 | 115 | 19.17 | |||
| 3 | 184 | 30.67 | |||
| 4 | 106 | 17.67 | |||
| 5 | 105 | 17.50 | |||
| Policy support recipiency (Sup_Y) | No (0) | 408 | 68.00 | ||
| Yes (1) | 192 | 32.00 | |||
| Research and development (Rnd_Y) | No (0) | 309 | 51.50 | ||
| Yes (1) | 291 | 48.50 | |||
Measurements of variables were collected by hiring a professional survey firm; company representatives were asked to answer the survey. The population size in the survey period of 27 days (February 1 through 27, 2020) was 7,933 companies (the size was reduced to 7,511 as of June 17, 2020). Among them, 22.14% (1,756 companies) were located in Chungcheong (Daejeon, Sejong, and North and South Chungcheong), 30.72% (2,437) in the southwest (Gwangju, North and South Jeolla, and Jeju), 15.83% (1,256) in Daegu–North Gyeongsang, 16.80% (1,333) in the southeast (Busan, Ulsan, and South Gyeongsang), and 14.43% (1,145) in Gangwon.
Thus, this study proportionally sampled a total of 600 companies: 21.17% (127), 31.67% (190), 15.00% (90), 16.00% (96), and 16.17% (97) from the five regions (as highlighted in Table 2).
| Region | Population | Sample |
|---|---|---|
| Chungcheong | 22.14% (1,756) | 21.17% (127) |
| Southwest | 30.72% (2,437) | 31.67% (190) |
| Daegu–North Gyeongsang | 15.83% (1,256) | 15.00% (90) |
| Southeast | 16.80% (1,333) | 16.00% (96) |
| Gangwon | 14.43% (1,145) | 16.17% (97) |
Figure 3 illustrates the spatial distribution of the sampled companies by locating their latitude and longitude coordinates; these were transformed using Python for street addresses in KICOX internal data. Points are differently colored to discern the cases in which multiple companies were recruited from one complex.
The analysis is also based on data from the National Spatial Data Infrastructure Portal and the Korean Industrial Complex Corporation. Among the objective variables, expressway interchanges, harbors, high schools, and colleges—all of which are objective indicators of the local environment—data is obtained via the National Spatial Data Infrastructure Portal and pre-processed with ArcGIS. Internal data from the Industrial Complex Corporation is provided for the lot sales and operation rate. Most of the input variables are continuous, including subjective qualities as measured on the Likert 5-point scale. Categorical variables are coded as shown in Table 1.
Table 1 presents the descriptive statistics of the sample and confirms its representativeness. This study attempts inferential statistics for which variations in most research variables are adequate. (Meanwhile, as a probability sample, its descriptive statistics are also meaningful per se: The sampling error is ±3.6%p at the 95% confidence level.) However, the number of harbors (Harb), lot sales rate (LotS), and operation rate (Oper) have limited variations in their original forms. Thus, the variables have been changed to dummies according to whether the complex has an accessible harbor (Harb_Y) and whether the lot sales and operation rates are at 100% (LotS_F and Oper_F).
Data ProcessingRegarding the analytical technique, for 2 response variables or components, the research model includes 22 components and 79 indicators (29 objective and 50 subjective) (see Figures 3–4). Then, (1) it may face multicollinearity among the indicators. Additionally, when psychometrics is used to combine the indicators into the components conceptually, they are not correlated but independent. So, (2) formative measurement should be employed rather than reflective measurement through factor analysis. The figure shows the formative measurement by arrows flowing from indicators to their respective components. In contrast, the reflective measurement is signified by arrows flowing from components to their indicators. A widely known formative component is sociodemographics because, for example, a correlation between gender and age is theoretically irrelevant. For regression or path analysis between components, the traditional (3) covariance-based structural equation modeling (CB-SEM) ideally requires 3 to 5 indicators per component; otherwise, the model is often unidentifiable. However, components in this model have 1–12 indicators each. Thus, this study selected partial least squares SEM (PLS-SEM) to address the above concerns.
Path coefficients and component (latent variable) scores are estimated iteratively according to a four-step procedure in PLS-SEM settings. The iteration stops when convergence is obtained; for examples, see Figure 4 below. One (1) is a path coefficient (inner weight) estimation; two (2) is an inside approximation, which computes proxies for all Yj-tilde with the weighted sum of its adjacent Yi scores; three (3) is an outer weight estimation (notably, multiple regression is used because all components are formative); and four (4) is an outside approximation. The updated wkj-tilde weights in the third step are used to update the Yjn component scores in the fourth outside approximation step (Sarstedt, Ringle et al., 2021).
Regarding significance testing of PLS-SEM—this study reports the coefficient significance at the 90% confidence level—bootstrapping of PLS-SEM was based on the sample size (600 cases) and a sufficiently large number of resampling (5,000 times) for the analytical stability. Table 3 and Table 4 first confirm the validity of the five indexed components. Specifically, the path from a component formed by multiple indicators to one represented by a single overall quality/satisfaction indicator (e.g., SET -> SET_G) was significant for all five cases. (All results, including path coefficients and their significance as well as R2, are the same in the employment and sales models.) In the order of R2 (and standardized coefficients), the qualities of the complex and supportive policy (R2 > 0.6) were more validly formed than those of the networking, infrastructure, and local environment.
| Standardized coef. | S.D. | S.E. | t | R2 | ||
|---|---|---|---|---|---|---|
| Path Coefficients | ||||||
| SET -> SET_G | 0.6663 | *** | 0.0297 | 0.0297 | 22.4449 | 0.4440 |
| COM -> COM_G | 0.7992 | *** | 0.0201 | 0.0201 | 39.7787 | 0.6387 |
| INF -> INF_G | 0.6760 | *** | 0.0281 | 0.0281 | 24.0939 | 0.4570 |
| NET -> NET_G | 0.7243 | *** | 0.0272 | 0.0272 | 26.5942 | 0.5246 |
| SUP -> SUP_G | 0.7915 | *** | 0.0185 | 0.0185 | 42.8677 | 0.6265 |
| SET_G -> EMPNU | -0.0164 | 0.0345 | 0.0345 | 0.4758 | 0.4360 | |
| COM_G -> EMPNU | 0.0501 | 0.0361 | 0.0361 | 1.3907 | ||
| INF_G -> EMPNU | 0.0683 | * | 0.0393 | 0.0393 | 1.7364 | |
| NET_G -> EMPNU | -0.0433 | 0.0392 | 0.0392 | 1.1065 | ||
| SUP_G -> EMPNU | 0.0298 | 0.0374 | 0.0374 | 0.7962 | ||
| LOC -> EMPNU | -0.0341 | 0.0561 | 0.0561 | 0.6089 | ||
| ACC -> EMPNU | 0.0344 | 0.0497 | 0.0497 | 0.6918 | ||
| HR -> EMPNU | 0.0174 | 0.0464 | 0.0464 | 0.3745 | ||
| CX_TIM -> EMPNU | -0.0477 | 0.0327 | 0.0327 | 1.4588 | ||
| CX_SPA -> EMPNU | -0.0443 | 0.0445 | 0.0445 | 0.9946 | ||
| CX_ACT -> EMPNU | 0.1809 | *** | 0.0474 | 0.0474 | 3.8191 | |
| CY_TIM -> EMPNU | -0.1675 | *** | 0.0417 | 0.0417 | 4.0165 | |
| CY_SPA -> EMPNU | 0.2548 | *** | 0.0447 | 0.0447 | 5.6996 | |
| CY_OWN -> EMPNU | 0.1986 | *** | 0.0358 | 0.0358 | 5.5452 | |
| CY_IND -> EMPNU | -0.1636 | 0.1022 | 0.1022 | 1.6009 | ||
| CY_SUP -> EMPNU | 0.0574 | * | 0.0306 | 0.0306 | 1.8793 | |
| CY_RND -> EMPNU | 0.1955 | *** | 0.0306 | 0.0306 | 6.3852 | |
| Outer Weights | ||||||
| Set_1 -> SET | 0.1211 | ** | 0.0603 | 0.0603 | 2.0095 | |
| Set_2 -> SET | 0.1656 | *** | 0.0534 | 0.0534 | 3.1016 | |
| Set_3 -> SET | 0.2555 | *** | 0.0511 | 0.0511 | 4.9985 | |
| Set_4 -> SET | 0.2603 | *** | 0.0521 | 0.0521 | 4.9985 | |
| Set_5 -> SET | 0.1900 | *** | 0.0561 | 0.0561 | 3.3849 | |
| Set_6 -> SET | 0.1833 | *** | 0.0606 | 0.0606 | 3.0259 | |
| Set_7 -> SET | 0.3202 | *** | 0.0577 | 0.0577 | 5.5502 | |
| Set_8 -> SET | 0.2849 | *** | 0.0660 | 0.0660 | 4.3147 | |
| Com_1R -> COM | 0.1205 | *** | 0.0386 | 0.0386 | 3.1209 | |
| Com_2R -> COM | 0.0060 | 0.0352 | 0.0352 | 0.1701 | ||
| Com_3R -> COM | 0.2499 | *** | 0.0395 | 0.0395 | 6.3296 | |
| Com_4R -> COM | 0.2386 | *** | 0.0546 | 0.0546 | 4.3655 | |
| Com_5R -> COM | 0.3918 | *** | 0.0543 | 0.0543 | 7.2119 | |
| Com_6R -> COM | 0.1948 | *** | 0.0471 | 0.0471 | 4.1368 | |
| Com_7R -> COM | 0.2309 | *** | 0.0448 | 0.0448 | 5.1531 | |
| Inf_1R -> INF | -0.0230 | 0.0570 | 0.0570 | 0.4034 | ||
| Inf_2R -> INF | 0.1445 | ** | 0.0659 | 0.0659 | 2.1937 | |
| Inf_3R -> INF | 0.1742 | *** | 0.0537 | 0.0537 | 3.2432 | |
| Inf_4R -> INF | 0.2653 | *** | 0.0678 | 0.0678 | 3.9128 | |
| Inf_5R -> INF | -0.2314 | *** | 0.0496 | 0.0496 | 4.6619 | |
| Inf_6R -> INF | 0.7377 | *** | 0.0515 | 0.0515 | 14.3375 | |
| Inf_7R -> INF | 0.2033 | *** | 0.0556 | 0.0556 | 3.6542 | |
| Inf_8R -> INF | 0.0316 | 0.0490 | 0.0490 | 0.6450 | ||
| Inf_9R -> INF | 0.0013 | 0.0498 | 0.0498 | 0.0263 | ||
| Net_1 -> NET | 0.0709 | 0.0441 | 0.0441 | 1.6094 | ||
| Net_2 -> NET | 0.1361 | *** | 0.0495 | 0.0495 | 2.7490 | |
| Net_3 -> NET | 0.1376 | *** | 0.0512 | 0.0512 | 2.6856 | |
| Net_4 -> NET | 0.2176 | *** | 0.0462 | 0.0462 | 4.7128 | |
| Net_5 -> NET | 0.1881 | ** | 0.0839 | 0.0839 | 2.2404 | |
| Net_6 -> NET | 0.1393 | * | 0.0784 | 0.0784 | 1.7754 | |
| Net_7 -> NET | 0.2815 | *** | 0.0715 | 0.0715 | 3.9386 | |
| Net_8 -> NET | 0.0574 | 0.0654 | 0.0654 | 0.8777 | ||
| Net_9 -> NET | 0.2656 | *** | 0.0536 | 0.0536 | 4.9534 | |
| Sup_1 -> SUP | 0.0741 | 0.0469 | 0.0469 | 1.5805 | ||
| Sup_2 -> SUP | 0.1851 | *** | 0.0501 | 0.0501 | 3.6952 | |
| Sup_3 -> SUP | 0.1460 | *** | 0.0506 | 0.0506 | 2.8860 | |
| Sup_4 -> SUP | 0.0747 | * | 0.0439 | 0.0439 | 1.7029 | |
| Sup_5 -> SUP | 0.3121 | *** | 0.0547 | 0.0547 | 5.7068 | |
| Sup_6 -> SUP | 0.1000 | ** | 0.0495 | 0.0495 | 2.0175 | |
| Sup_7 -> SUP | -0.0146 | 0.0471 | 0.0471 | 0.3087 | ||
| Sup_8 -> SUP | 0.1393 | *** | 0.0423 | 0.0423 | 3.2901 | |
| Sup_9 -> SUP | 0.0065 | 0.0422 | 0.0422 | 0.1537 | ||
| Sup_10 -> SUP | 0.1244 | ** | 0.0493 | 0.0493 | 2.5243 | |
| Sup_11 -> SUP | 0.0955 | ** | 0.0439 | 0.0439 | 2.1751 | |
| Sup_12 -> SUP | 0.3136 | *** | 0.0428 | 0.0428 | 7.3208 | |
| Set_G -> SET_G | 1 | 0 | 0 | 0 | ||
| Com_G -> COM_G | 1 | 0 | 0 | 0 | ||
| Inf_G -> INF_G | 1 | 0 | 0 | 0 | ||
| Net_G -> NET_G | 1 | 0 | 0 | 0 | ||
| Sup_G -> SUP_G | 1 | 0 | 0 | 0 | ||
| LocRD1 -> LOC | 0.3187 | 0.2999 | 0.2999 | 1.0628 | ||
| LocRD2 -> LOC | -0.1885 | 0.3106 | 0.3106 | 0.6070 | ||
| LocRD3 -> LOC | 0.8833 | 0.7661 | 0.7661 | 1.1531 | ||
| LocRD5 -> LOC | -0.1491 | 0.2772 | 0.2772 | 0.5378 | ||
| Expway -> ACC | 0.6520 | 0.6298 | 0.6298 | 1.0353 | ||
| Harb_Y -> ACC | -0.6425 | 0.6396 | 0.6396 | 1.0046 | ||
| Sch_Co -> HR | -0.6645 | 0.7966 | 0.7966 | 0.8342 | ||
| Sch_Hi -> HR | 1.2000 | 0.7955 | 0.7955 | 1.5085 | ||
| Yr_Cx -> CX_TIM | 1 | 0 | 0 | 0 | ||
| Are_As -> CX_SPA | 0.7472 | 2.2222 | 2.2222 | 0.3363 | ||
| Are_Av -> CX_SPA | 0.9627 | *** | 0.1108 | 0.1108 | 8.6899 | |
| Are_Ma -> CX_SPA | -1.9447 | 1.7757 | 1.7757 | 1.0952 | ||
| Are_To -> CX_SPA | 1.3242 | * | 0.7931 | 0.7931 | 1.6697 | |
| Emp_Av -> CX_ACT | 1.1369 | *** | 0.0532 | 0.0532 | 21.3625 | |
| Exp_Av -> CX_ACT | -0.2259 | ** | 0.1091 | 0.1091 | 2.0716 | |
| LotS_F -> CX_ACT | 0.0625 | 0.0770 | 0.0770 | 0.8115 | ||
| Oper_F -> CX_ACT | -0.1872 | ** | 0.0929 | 0.0929 | 2.0154 | |
| Sal_Av -> CX_ACT | -0.1844 | 0.1122 | 0.1122 | 1.6428 | ||
| Yr_CyE -> CY_TIM | 0.8808 | *** | 0.1336 | 0.1336 | 6.5931 | |
| Yr_CyM -> CY_TIM | 0.1677 | 0.1746 | 0.1746 | 0.9600 | ||
| Are_Bu -> CY_SPA | -3.8826 | 2.8256 | 2.8256 | 1.3741 | ||
| Are_Si -> CY_SPA | 4.2034 | * | 2.4924 | 2.4924 | 1.6865 | |
| Own_Y -> CY_OWN | 1 | 0 | 0 | 0 | ||
| InduRD2 -> CY_IND | 0.2175 | 0.1820 | 0.1820 | 1.1955 | ||
| InduRD3 -> CY_IND | 0.7662 | * | 0.4009 | 0.4009 | 1.9114 | |
| InduRD4 -> CY_IND | -0.3327 | 0.3282 | 0.3282 | 1.0135 | ||
| InduRD5 -> CY_IND | 0.6798 | * | 0.3547 | 0.3547 | 1.9165 | |
| Sup_Y -> CY_SUP | 1 | 0 | 0 | 0 | ||
| Rnd_Y -> CY_RND | 1 | 0 | 0 | 0 | ||
| Ln_EmpNu -> EMPNU | 1 | 0 | 0 | 0 | ||
* p < 0.1, ** p < 0.05, *** p < 0.01
Note: For variable names, see Table 1.
| Standardized coef. | S.D. | S.E. | t | R2 | ||
|---|---|---|---|---|---|---|
| Path Coefficients | ||||||
| SET -> SET_G | 0.6663 | *** | 0.0300 | 0.0300 | 22.2253 | 0.4440 |
| COM -> COM_G | 0.7992 | *** | 0.0200 | 0.0200 | 39.9637 | 0.6387 |
| INF -> INF_G | 0.6760 | *** | 0.0281 | 0.0281 | 24.0221 | 0.4570 |
| NET -> NET_G | 0.7243 | *** | 0.0269 | 0.0269 | 26.8852 | 0.5246 |
| SUP -> SUP_G | 0.7915 | *** | 0.0184 | 0.0184 | 43.0073 | 0.6265 |
| SET_G -> TSAL9 | 0.0162 | 0.0379 | 0.0379 | 0.4267 | 0.4028 | |
| COM_G -> TSAL9 | 0.0224 | 0.0375 | 0.0375 | 0.5962 | ||
| INF_G -> TSAL9 | -0.0112 | 0.0370 | 0.0370 | 0.3022 | ||
| NET_G -> TSAL9 | -0.0108 | 0.0396 | 0.0396 | 0.2731 | ||
| SUP_G -> TSAL9 | 0.0712 | ** | 0.0358 | 0.0358 | 1.9909 | |
| LOC -> TSAL9 | -0.0931 | 0.0983 | 0.0983 | 0.9478 | ||
| ACC -> TSAL9 | 0.0450 | 0.0515 | 0.0515 | 0.8736 | ||
| HR -> TSAL9 | 0.0666 | 0.0674 | 0.0674 | 0.9881 | ||
| CX_TIM -> TSAL9 | -0.0052 | 0.0297 | 0.0297 | 0.1738 | ||
| CX_SPA -> TSAL9 | 0.0521 | 0.0455 | 0.0455 | 1.1450 | ||
| CX_ACT -> TSAL9 | 0.1276 | *** | 0.0473 | 0.0473 | 2.6983 | |
| CY_TIM -> TSAL9 | -0.1953 | *** | 0.0397 | 0.0397 | 4.9159 | |
| CY_SPA -> TSAL9 | 0.1755 | *** | 0.0438 | 0.0438 | 4.0036 | |
| CY_OWN -> TSAL9 | 0.2184 | *** | 0.0377 | 0.0377 | 5.7997 | |
| CY_IND -> TSAL9 | 0.0272 | 0.0450 | 0.0450 | 0.6042 | ||
| CY_SUP -> TSAL9 | 0.0538 | * | 0.0300 | 0.0300 | 1.7943 | |
| CY_RND -> TSAL9 | 0.1899 | *** | 0.0320 | 0.0320 | 5.9288 | |
| Outer Weights | ||||||
| Set_1 -> SET | 0.1211 | ** | 0.0591 | 0.0591 | 2.0482 | |
| Set_2 -> SET | 0.1656 | *** | 0.0528 | 0.0528 | 3.1356 | |
| Set_3 -> SET | 0.2555 | *** | 0.0511 | 0.0511 | 5.0052 | |
| Set_4 -> SET | 0.2603 | *** | 0.0504 | 0.0504 | 5.1602 | |
| Set_5 -> SET | 0.1900 | *** | 0.0562 | 0.0562 | 3.3830 | |
| Set_6 -> SET | 0.1833 | *** | 0.0596 | 0.0596 | 3.0735 | |
| Set_7 -> SET | 0.3202 | *** | 0.0572 | 0.0572 | 5.5925 | |
| Set_8 -> SET | 0.2849 | *** | 0.0638 | 0.0638 | 4.4649 | |
| Com_1R -> COM | 0.1205 | *** | 0.0380 | 0.0380 | 3.1699 | |
| Com_2R -> COM | 0.0060 | 0.0353 | 0.0353 | 0.1696 | ||
| Com_3R -> COM | 0.2499 | *** | 0.0383 | 0.0383 | 6.5220 | |
| Com_4R -> COM | 0.2386 | *** | 0.0539 | 0.0539 | 4.4227 | |
| Com_5R -> COM | 0.3918 | *** | 0.0540 | 0.0540 | 7.2503 | |
| Com_6R -> COM | 0.1948 | *** | 0.0464 | 0.0464 | 4.1946 | |
| Com_7R -> COM | 0.2309 | *** | 0.0443 | 0.0443 | 5.2143 | |
| Inf_1R -> INF | -0.0230 | 0.0572 | 0.0572 | 0.4024 | ||
| Inf_2R -> INF | 0.1445 | ** | 0.0648 | 0.0648 | 2.2309 | |
| Inf_3R -> INF | 0.1742 | *** | 0.0544 | 0.0544 | 3.2046 | |
| Inf_4R -> INF | 0.2653 | *** | 0.0674 | 0.0674 | 3.9350 | |
| Inf_5R -> INF | -0.2314 | *** | 0.0496 | 0.0496 | 4.6664 | |
| Inf_6R -> INF | 0.7377 | *** | 0.0515 | 0.0515 | 14.3256 | |
| Inf_7R -> INF | 0.2033 | *** | 0.0564 | 0.0564 | 3.6066 | |
| Inf_8R -> INF | 0.0316 | 0.0503 | 0.0503 | 0.6274 | ||
| Inf_9R -> INF | 0.0013 | 0.0507 | 0.0507 | 0.0258 | ||
| Net_1 -> NET | 0.0709 | 0.0442 | 0.0442 | 1.6057 | ||
| Net_2 -> NET | 0.1361 | *** | 0.0492 | 0.0492 | 2.7685 | |
| Net_3 -> NET | 0.1376 | *** | 0.0512 | 0.0512 | 2.6863 | |
| Net_4 -> NET | 0.2176 | *** | 0.0469 | 0.0469 | 4.6369 | |
| Net_5 -> NET | 0.1881 | ** | 0.0837 | 0.0837 | 2.2473 | |
| Net_6 -> NET | 0.1393 | * | 0.0776 | 0.0776 | 1.7951 | |
| Net_7 -> NET | 0.2815 | *** | 0.0701 | 0.0701 | 4.0181 | |
| Net_8 -> NET | 0.0574 | 0.0651 | 0.0651 | 0.8817 | ||
| Net_9 -> NET | 0.2656 | *** | 0.0543 | 0.0543 | 4.8945 | |
| Sup_1 -> SUP | 0.0741 | 0.0470 | 0.0470 | 1.5766 | ||
| Sup_2 -> SUP | 0.1851 | *** | 0.0505 | 0.0505 | 3.6643 | |
| Sup_3 -> SUP | 0.1460 | *** | 0.0499 | 0.0499 | 2.9248 | |
| Sup_4 -> SUP | 0.0747 | * | 0.0442 | 0.0442 | 1.6908 | |
| Sup_5 -> SUP | 0.3121 | *** | 0.0542 | 0.0542 | 5.7578 | |
| Sup_6 -> SUP | 0.1000 | ** | 0.0484 | 0.0484 | 2.0656 | |
| Sup_7 -> SUP | -0.0146 | 0.0461 | 0.0461 | 0.3154 | ||
| Sup_8 -> SUP | 0.1393 | *** | 0.0425 | 0.0425 | 3.2751 | |
| Sup_9 -> SUP | 0.0065 | 0.0419 | 0.0419 | 0.1547 | ||
| Sup_10 -> SUP | 0.1244 | ** | 0.0490 | 0.0490 | 2.5362 | |
| Sup_11 -> SUP | 0.0955 | ** | 0.0440 | 0.0440 | 2.1705 | |
| Sup_12 -> SUP | 0.3136 | *** | 0.0424 | 0.0424 | 7.3984 | |
| Set_G -> SET_G | 1 | 0 | 0 | 0 | ||
| Com_G -> COM_G | 1 | 0 | 0 | 0 | ||
| Inf_G -> INF_G | 1 | 0 | 0 | 0 | ||
| Net_G -> NET_G | 1 | 0 | 0 | 0 | ||
| Sup_G -> SUP_G | 1 | 0 | 0 | 0 | ||
| LocRD1 -> LOC | 0.6518 | 0.5349 | 0.5349 | 1.2185 | ||
| LocRD2 -> LOC | -0.3191 | 0.3720 | 0.3720 | 0.8578 | ||
| LocRD3 -> LOC | 0.5241 | 0.4413 | 0.4413 | 1.1876 | ||
| LocRD5 -> LOC | -0.2700 | 0.3255 | 0.3255 | 0.8296 | ||
| Expway -> ACC | 0.7891 | 0.5555 | 0.5555 | 1.4205 | ||
| Harb_Y -> ACC | -0.4803 | 0.4904 | 0.4904 | 0.9795 | ||
| Sch_Co -> HR | -0.8900 | 0.8580 | 0.8580 | 1.0373 | ||
| Sch_Hi -> HR | 1.1628 | 0.8719 | 0.8719 | 1.3337 | ||
| Yr_Cx -> CX_TIM | 1 | 0 | 0 | 0 | ||
| Are_As -> CX_SPA | 0.0672 | 2.1666 | 2.1666 | 0.0310 | ||
| Are_Av -> CX_SPA | 0.8806 | *** | 0.0747 | 0.0747 | 11.7907 | |
| Are_Ma -> CX_SPA | -1.3935 | 1.8584 | 1.8584 | 0.7498 | ||
| Are_To -> CX_SPA | 1.6829 | *** | 0.6437 | 0.6437 | 2.6144 | |
| Emp_Av -> CX_ACT | 1.0581 | *** | 0.0683 | 0.0683 | 15.4889 | |
| Exp_Av -> CX_ACT | -0.2126 | * | 0.1163 | 0.1163 | 1.8283 | |
| LotS_F -> CX_ACT | 0.1495 | 0.1026 | 0.1026 | 1.4569 | ||
| Oper_F -> CX_ACT | -0.1427 | 0.1010 | 0.1010 | 1.4122 | ||
| Sal_Av -> CX_ACT | 0.0165 | 0.0990 | 0.0990 | 0.1664 | ||
| Yr_CyE -> CY_TIM | 0.9306 | *** | 0.0823 | 0.0823 | 11.3011 | |
| Yr_CyM -> CY_TIM | 0.1003 | 0.1134 | 0.1134 | 0.8849 | ||
| Are_Bu -> CY_SPA | -3.8342 | 2.7766 | 2.7766 | 1.3809 | ||
| Are_Si -> CY_SPA | 4.1794 | * | 2.4516 | 2.4516 | 1.7048 | |
| Own_Y -> CY_OWN | 1 | 0 | 0 | 0 | ||
| InduRD2 -> CY_IND | 0.4801 | 0.3228 | 0.3228 | 1.4874 | ||
| InduRD3 -> CY_IND | -0.2206 | 0.4044 | 0.4044 | 0.5454 | ||
| InduRD4 -> CY_IND | 0.7615 | * | 0.4331 | 0.4331 | 1.7584 | |
| InduRD5 -> CY_IND | -0.2156 | 0.3592 | 0.3592 | 0.6001 | ||
| Sup_Y -> CY_SUP | 1 | 0 | 0 | 0 | ||
| Rnd_Y -> CY_RND | 1 | 0 | 0 | 0 | ||
| Ln_TSal9 -> TSAL9 | 1 | 0 | 0 | 0 | ||
* p < 0.1, ** p < 0.05, *** p < 0.01
Note: For variable names, see Table 1.
Concerning psychometrics, Table 3 and Table 4 show which one represents its component (the following results are the same in the two models). While all the eight environmental indicators (beginning with “Set_”) significantly formed the local environment component (SET), among the seven complex indicators (Com_), the second (Com_2R), “[a]gglomeration effects occur because companies in specific industries are concentrated,” was insignificant. This may imply that company representatives have a low expectation of agglomeration since a reciprocal relationship between companies has not been emphasized from the initial stage of the agro-industrial complex development (Kim, S.-B. and Hong, 2010; Lim, An et al., 2010) .
The overall satisfaction/quality of the infrastructure was not affected by the qualities of waste disposal systems (Inf_1R) and (community and residential) facilities for employees (Inf_8R–Inf_9R). Instead, the other significant indicators were directly associated with sales, such as production, logistics (storage), and transportation. This result can be partially attributed to the fact that survey respondents were company representatives, not employees.
Regarding the networking environment, academic/research institutions (Net_1) and central ministries (Net_8) were not emphasized compared to residents, complex/local companies, and local/provincial governments. That is, companies favored local networks over national ones.
Lastly, insignificant among supportive policy indicators were grants (Sup_1) and administrative services (Sup_) for the complex location/investment and support for building cooperative relationships with business, academic, and research institutions (Sup_7). The other significant indicators were somewhat related to continuous supports in contrast to these one/short-time supports.
Variable significanceTable 3 and Table 4 show that the variation explained by the direct paths of the 17 components was similar in the employment (R2= 0.4360) and sales (0.4028) models. Also, according to the significance (unlike the magnitude to be discussed later), the two models led to the same results in terms of the 12 objective condition components (in Figure 5 and Figure 6 the left parts of the response component); the results differed regarding the significance of the five subjective quality components. Notably, half of the 12 components were insignificant in the employment and sales models: as regional-scale components, accessibility (ACC), human resources (HR), and location (LOC), and on the complex scale, space (represented by the designated year) (CX_TIM) and time (four indicators) (CX_SPA) characteristics, and among company-scale components, industry type (CY_IND).
Transportation (accessibility) and labor (human resources) were considered to control for other traditional industry location factors to an extent; location and industry type were insignificant. This is likely due to how the site of a complex company has often been arbitrarily determined on psychological and cultural grounds, not by objective and impersonal analyses, as discussed in “[t]hreats to the competitiveness of agro-industrial complexes.” The insignificance of the designated year, as a representative of the temporal characteristics of the complex (Lee, K.-R., 2015), echoes Lee’s finding (Lee, K.-R., 2015) and rejects the suspicion of previous studies (Choi, K., Kim, Y. et al., 2012; Jang, C.-S. , 2015; Kim, S.-B. and Hong, 2010) that the age of the complex would lower the complex vitality. Also, while previous studies (Choi, K., Kim, Y. et al., 2012; Kim, S.-B. and Hong, 2010) considered that small-scale development negatively influences complex competitiveness and Lee, K.-R. (2015) empirically confirmed that finding, this study found any influence insignificant. The inconsistency may be explained by the fact that Lee studied only one province (not the entire country) and used the entire complex (not its company) as the unit of analysis. All in all, significant variables mainly were company characteristics (beginning with “CY_”)—only industry type (CY_IND) was not significant—and among complex characteristics, only activity level (CX_ACT) was statistically significant.
[Standardized coefficients (t-values)]
Note: The above figures were created using PLS-SEM freeware, SmartPLS 2.0.M3. For variable names, see Table 1.
[Standardized coefficients (t-values)]
Note: The above figures were created using PLS-SEM freeware, SmartPLS 2.0.M3. For variable names, see Table 1.
Among the significant objective components, only the temporal characteristics of the company (CY_TIM) had negative coefficients in both models. As in Figures 3–4, between the two indicators of the component, established (Yr_CyE) and moved (Yr_CyM) years, only the former was significant. Thus, the negative direction suggests that (regardless of when it was moved into the complex) a younger company has fewer employees and lower sales.
As entirely consistent in the significance and direction, the objective components were only partially similar in the relative magnitude between the employment and sales models. Among those with positive coefficients, policy support recipiency (CY_SUP) and complex activity level (CX_ACT) had the lowest magnitudes in the two models, implying that the employment size and sales are less concerned with complex-level variables than company-level variables (variations in policy support is also meaningful between complexes). Out of the two minor components, the effect of policy support had a lower magnitude: compared to other objective components, 22.53% (= 0.0574/0.2548)–31.73% (= 0.0574/0.1809) in the employment model and 24.63% (= 0.0538/0.2184)–42.16% (= 0.0538/0.1276) in the sales model. Notably, on sales, the objective component of policy support exerted a weaker influence than the subjective component of policy quality (to be discussed later). Lastly, employment had an overwhelming effect among the five indicators of the complex activity level—full lot sales and operation, average employment, production, and exports. Other significant indicators (average exports in the employment model and average exports and full operation in the sales model) had adverse effects. This result denotes that if neighboring companies in the same complex hire employees and focus on the domestic market, it will positively affect the employment and sales of the company; if their operation is currently below 100%, sales will further increase from these measures.
The other three objective components had differing magnitudes in the two models. These components are company spatial characteristics (of the two indicators of this component, site area and building area, only the former showed significance in both models), company ownership, and research and development (R&D). In descending order, spatial characteristics, ownership, and R&D in the employment model. In the sales model, they are ownership, R&D, and spatial characteristics; however, their effects were more significant than those from the above-discussed policy support and complex activity level. A company located on a large site with ownership will likely have more employees and sales. Also, although companies are based on rural labor force and primary industries, innovations through R&D can positively influence employment (e.g., research staff hiring) and sales (e.g., value-added), supporting the argument of Lee, C.-W. (2008) and Woo, J.M. (2008). In fact, technological advancements through R&D and its financing have functioned as a viable sustainability strategy in countries with large-scale agro-industrial complexes, such as Russia (Sergi, Popkova et al., 2019). Additionally, (Liu, Wang et al., 2021) argued that technological innovation helps increase the market’s competitiveness and future potential, segmentation, and diversification.
The most notable result is that among the five subjective quality components, and based on 45 indicators, the employment size was affected by the infrastructure quality in the employment model and the sales by the quality of supportive policy in the sales model. Notably, in the sales model, the coefficient was 0.0712 (SUP_G -> TSAL9) and somewhat more significant than that of the objective status of policy support recipiency: 0.0538 (CY_SUP -> TSAL9). Below, this study discusses the implications of these subjective components.
Employment and the economy are two major focuses of Korean regional policy (You and Byun, 2011). At its initial stage, the agro-industrial complex attempted to use off-season human resources in rural areas and advance its industrial structure. Although agro-industrial complexes were found to be located in poor environments with disadvantages from the perspective of location theories, they were still beneficial in attracting the population through new hiring and revitalizing the depressed rural economy (Gim, Choi et al., 2020).
First, concerning employment size, a company can attract employees by improving its infrastructure (INF_G -> EMPNU). However, building new community facilities for employees (Inf_8R) and residential facilities (e.g., dorms) (Inf_9R) is unlikely to be attractive or significant. Instead, road quality (Inf_6R) was overwhelmingly important: The weight (0.7377) was 2.78–5.11 times larger than other indicators. Furthermore, whether parking space is sufficient (Inf_7R) was more important than the gas and water supply stability. This implies that factory systems equipped with employee dormitories—common before the 1990s—may not work in a contemporary setting. Instead, a new lifestyle needs to be acknowledged for employees to commute and spend time for leisure, shopping, and personal affairs outside the complex. Notably, whether logistics facilities are sufficient (Inf_5R) had a negative coefficient; only this indicator worked unattractively on potential employees. The result is consistent with the finding that logistics facilities aggravate the regional image (Jin and Paulsen, 2018).
Second, while efforts continue for more sales, such as facility improvement, this study found that indirect support for the complex’s interior and exterior environments, infrastructure, and networking is insignificant, and only direct support is effective (SUP_G -> TSAL9). This confirms Lee’s suspicion (Lee, K.-R., 2015) that agro-industrial complexes may not be vitalized without physical improvements. Still, Kim and Kang (2011), in their study on industrial complexes eligible for the Industrial Complex Regeneration program, argued that a program is ineffective if focused on only physical (e.g., infrastructure) improvements; instead, they valued a program that utilized accounting, financing, and community services.
Meanwhile, regarding specific types of direct support, tax/financial benefits (Sup_2) were less important than the active response of the government (Sup_12). This implies a pressing need for organized support in as much as the Central Association for Agro-Industrial Complexes has no administrative headquarters to hear the individual member companies’ voices and conduct independent projects. As for the levels of support institutions, support at the provincial level (Sup_5) outweighed district-level support (Sup_3). It may suggest that among other direct programs, options that are more directly related to sales should be favored: According to Shin (2014), to support small enterprises, provincial governments concentrate more on marketing and market expansion while the central government focuses on R&D and HRD and local/district governments on the complex designation and management. Additionally, Liu and Li (2021) argued that the impact of R&D and technological innovation was more dominant in less developed cities.
This study contributes to the literature by examining decisive factors affecting the employment and sales of agro-industrial complexes for regional revitalization, according to location theory, based on empirical analysis of industrial space production efficiency and sustainability.
Conceptualization, T-H.T.G., J.L. and J.C.; methodology, T-H.T.G., J.L. and J.C.; software, T-H.T.G.; investigation, T-H.T.G. and J.L.; resources, J.L.; data curation, T-H.T.G. and J.L.; writing—original draft preparation, T-H.T.G., J.L. and J.C.; writing—review and editing, T-H.T.G., J.L. and J.C.; supervision, T-H.T.G. and J.C. All authors have read and agreed to the published version of the manuscript.
The authors declare that they have no conflicts of interest regarding the publication of the paper.
This research was funded by the Ministry of Education of the Republic of Korea and the National Research Foundation of Korea(NRF-2021S1A3A2A01087370).
This research was supported by the Ministry of Education of the Republic of Korea and the National Research Foundation of Korea (NRF-2021S1A3A2A01087370).
