台灣製造業都市聚集經濟之研究

An Investigation of Urban Agglomeration Economies in Taiwanese Manufacturing

都市聚集經濟被認為是都市形成的主要原因之一，但過去研究聚集經濟時，大多利用特定的生產函數來作實證估計，而資料包絡分析法（DEA）則不受這樣的限制，因此本文利用此法來估計台灣各都市之製造業的生產效率分數。聚集經濟一般被區分為都市化經濟與地方化經濟，後者過去多使用產業的就業人數來代表，但本文以為相對的都市產業產出比例應是更好的選擇，這樣不但可以解決統計上的直線性重合問題，更可引用區位商數來加以支持。在利用效率分數與聚集經濟進行截斷Tobit迴歸分析後發現，大多數的中分類製造業的生產效率在1996至2001年所依賴的是地方化經濟而非都市化經濟，但聚集經濟的影響力有下降的趨勢。 The most fundamental topic in urban economics is why cites form, and the most popular explanation is agglomeration economies. In the past, empirical studies have mostly set an a priori specific production function and used it to estimate urban productivity. This paper proposes Data Envelopment Analysis (DEA) to avoid this problem and then computes urban efficiency scores. Agglomeration economies are generally divided into urbanization economies and localization economies. The latter was usually estimated using the industrial employment, but it seems more rational to use an industrial output ratio, which not only reduces the overly high correlation between explanatory variables, but also introduces the support for a location quotient (LQ). Regressing efficiency scores on agglomeration economies by censored Tobit method reveals that many industries mainly depend on localization economies, but agglomeration effects fell gradually during 1996 and 2001.