تجزیه و تحلیل هم انباشتگی و اثرات جانبی پویا

This paper presents a cointegration analysis on the effects of dynamic externalities upon economic growth using time-series data from 1975 to 2003 on the one-digit industries of the Tokyo metropolitan area in Japan. Some new time-series econometric methods that have been recently developed to conduct unit root and cointegration tests are used in the analysis, allowing for an endogenously determined structural change in the time period studied. It also proposes a new type of dynamic externalities, called Network dynamic externalities, to represent knowledge spillovers resulting from the whole agglomerated area via transportation networks, and shows that they have cointegrated relations with the total factor productivity (TFP) of the manufacturing, finance, wholesale and retail trade, as well as the overall industries. In addition, evidence is also found that Marshall–Arrow–Romer (MAR) dynamic externalities, which are associated with own industrial production concentration, affect the TFP of most industries selected for estimation. However, Jacobs dynamic externalities, which are represented by the diversity of industrial production, only contribute to the TFP of the services industry, and Porter dynamic externalities, which are expressed by the competitiveness within industries, do not influence the selected industrial TFP.

Recent endogenous economic growth theories (Romer, 1986 and Lucas, 1988) argue that economic growth can result endogenously from technological progress, which is brought about mainly through knowledge spillovers. Such knowledge spillovers are most likely to occur when firms of one or different industries agglomerate to a limited urban area. The resulting agglomeration economies are called dynamic externalities, which are considered as an engine of economic growth (see Glaeser et al., 1992 and Henderson et al., 1995).

This paper presented a cointegration analysis of the effects that dynamic externalities exert upon economic growth, using time-series data from 1975 to 2003 on the one-digit industries of the Tokyo metropolitan area in Japan. In the analysis, we applied some new methods that have been recently developed in time-series econometrics to carry out unit root and cointegration tests, allowing for an endogenously determined structural change within the time period for study. Concerning the contents of dynamic externalities, we discussed the limitation of the existing classification that divides them into three types, i.e., the types of Marshall–Arrow–Romer (MAR), Jacobs and Porter. We proposed a new type of dynamic externalities to represent those externalities that occur from the interactions among not only the firms of industries but also among households outside the industries and between those firms and households as well. We called this new type Network dynamic externalities, emphasizing that they are related to all the networks of human beings, and represented it by using the density of transportation networks within the whole study area in question.