![Sample our Urban Studies journals, sign in here to start your FREE access for 14 days]({"type":"image" , "src" : "/sda/110826/TOC-Subject-Sample-2021-Urban-Studies.jpg"})
Abstract
It is well acknowledged that agglomeration economies contribute to knowledge production, which is crucial in the current knowledge-based economy. The literature on evolutionary agglomeration stresses that the effects of agglomeration economies may vary depending on the industry life cycle. But much is not known about what sources of agglomeration economies play an important role in knowledge production at each stage of the life cycle: industrial localization versus diversification, and leadership by large firms versus competition among small firms. This study investigates knowledge production in the Information and Communication Technology (ICT) industry in the capital area around Seoul, South Korea, to answer this question. A panel data set for the patent applications during the period of 2006–2011 is used to model knowledge production in terms of different sources of agglomeration economies. The results show that distinct patterns can be found depending on the industry life cycle. Leadership by large firms has positive effects on knowledge production during the growth stage (2006–2009), while its effects become insignificant during the maturity stage (2009–2011). This is in line with the expectations of evolutionary agglomeration theory. The effects of industrial diversity change from negative to insignificant as the ICT industry enters the maturity stage. In the meantime, competition always has positive effects, while localization always has no significant effects.
ORCID
In Kwon Park http://orcid.org/0000-0001-7043-0485
Notes
2. The SCA is the metropolitan area of the capital, consisting of Seoul, Incheon, and Gyeonggi-do. The spatial unit of analysis is the municipality, Si-Gun-Gu, and the SCA includes 66 municipalities. From 2006 to 2011, the number of patent applications in the ICT industry has increased in the SCA and accounts for about 80% of the total patent applications in the country: 88% in 2006, 88% in 2007, 83% in 2008, 78% in 2009, 80% in 2010, and 80% in 2011.
3. This way of identifying the place of research may not be perfect, as the applicant may report the wrong place, for instance the address of the headquarters. But according to an official who manages the SAPI with the Korean Intellectual Property Office, this is not a common case in the application process. While the locations of the headquarters are sometimes reported for registering patents, the actual places of invention tend to be reported for applying for them. This is why many empirical studies use the same method of place identification; for instance, see Lee and Lee (Citation2010).
4. We also tried a model regarding the level of diversity within the ICT industry to see the effect of relatedness on knowledge production as suggested by Boschma et al. (Citation2015). However, the results show that the variable is not significant at all, and thus it has been omitted from the final model.
5. The ICT-related services are defined in terms of KSIC to include J582 (software development and supply), J591 (movies, videos, and TV programmes production and distribution), J592 (audios publication and original recording), J601 (radio broadcasting), J602 (TV broadcasting), J620 (computer programming, system integration and management), J631 (data processing, hosting, portal, and other Internet media services), and J639 (other information services).
6. The command of xtgls in STATA 12 was used. With no options, xtgls supposes that all error terms are independent and identically distributed. If there are auto-correlation and heteroscedasticity of the error term, corr (ar1) and panel (hetrr) and corr (ar1) help one solve these problems, respectively.
7. The omitted four municipalities are Yangpyeong-gun, Yeoncheon-gun, Ganhwa-gun, and Ongjin-gun.