| 摘要: | 文獻研究指出一國區域勞動市場存在聚集現象。台灣整體失業率在2000年開始大幅攀升，許多研究針對影響總體失業攀升的因素作討論，然而卻鮮少研究針對區域失業率多加著墨。此外，台灣區域失業率的變異數從1988年後逐漸降低，然文獻上對此問題的研究卻付之闕如。為此，本研究探討兩個主要問題：一為討論影響台灣區域失業率的因素為何？台灣區域失業率是否亦存在空間聚集現象？二為討論影響區域失業率變異數的可能因素為何？ 首先在第一個問題上，本研究建構台灣22縣市（扣除離島澎湖、金門、連江三縣）1982-2005年的縱橫資料集，利用空間相依模型（Spatial Autoregressive Model），以區域工資作為工具變數並採兩階段工具變數法進行估計。與過去文獻不同的是，本研究同時考量空間外溢、區域特性與一國總體環境等三類型因素對區域失業的影響；且在空間外溢影響上，除如文獻納入第一階區域失業落遲因子（相鄰區域平均失業率， ）外，本研究首次納入第二階區域失業落遲因子（ ）探討空間外溢與距離間的關係。 經檢定後發現，放入第二階區域失業落遲因子可有效排除縱橫資料集所產生的空間相依性問題。實證研究結果發現，縣市勞動特性、總體經濟以及空間外溢等三類型因素確實同時對台灣區域失業率產生影響。值得重視的是，納入兩個空間落遲失業率因子( 、 )後，雖然區域勞動特性與總體經濟指標對台灣區域失業率仍具影響力，但空間外溢效果較可能為影響台灣區域失業率的主要因素。 與 的符號皆為正且 的估計係數高於 的實證結果，不僅證實了台灣縣市勞動市場確實存在空間聚集現象，且此外溢擴散程度亦隨距離之拉長而遞減。納入空間落遲失業率因子後，代表人力資本的相關變數其估計結果皆呈不顯著，顯示人力資本並非為影響台灣各縣市失業率的關鍵因素。 此外本研究亦發現，都市化程度較高、行政資源較為充裕的直轄市，其勞動市場顯著受到相鄰縣份的正向外溢影響，代表包圍著直轄市之縣份的勞動市場表現優劣，將對被其包圍的直轄市之勞動此市場產生極大影響。再者，本研究將區域重新劃分成15區，將七個直轄市與包圍著該直轄市之縣市視為同一勞動市場。比較22區域與15區域的實證結果顯示，當擴大都會區的勞動市場範圍時，區域空間影響效果將變得更強，而此亦可證實台灣勞動市場在地理分布上有相當的聚集。 值得一提的是，本研究為首次在文獻上針對影響區域失業變異數的可能因素作探討。研究結果顯示，雖然如前述的實證結果顯示人力資本對各區域失業率不具影響力，惟本部份的實證結果證實勞動力品質的提升確可明顯降低台灣區域失業的差異；另當區域間的產業比重、工資水準差異越大時，區域失業率的變異數也越高，特別是製造業產業比重的差異其對區域變異數的影響非常顯著且亦十分穩定。此外，不同交通設施對區域失業差異的影響不盡相同。長距離的鐵路運輸所花費的時間成本較高對區域失業變異則將會產生正向影響，相反地，花費時間較少的航空運輸則將會降低區域失業的變異。然而，本研究的實證結果顯示，高速公路等路運交通的改善可能反倒使得區域勞動市場更加聚集，其對改善城鄉差距的效果反倒不十分明確。 Many studies suggest that it appears that unemployment tends to cluster in regional labor markets. Reviewing literatures about Taiwan’s regional unemployment, such a spatial effect has not been examined intensively. Therefore the two main questions that this dissertation shall seek to answer are: First, what are the determinants in regional unemployment rates? Does there exist a positive spatial spillover in the regional labor market of Taiwan? Second, what are possible influences to the variance of regional unemployment rates? To facilitate such a study, a panel dataset of 22 regions in Taiwan from 1982 to 2005 was compiled for the first issue. Spatial autoregressive modeling with instruments to examine the possible factors to regional unemployment rates is adopted. Differing from the current literature, regional-specific characteristics, the macro-environment and spatial spillovers are simultaneously considered in discussing the determinants to regional unemployment rates. Besides the first–order lag variable of the spatially-weighted average unemployment rates which is employed in the current literature, this study constructed a second-order lag variable of spatially-weighted average unemployment rates to specifically account for the impact of distances for the first time. It is confirmed that spatial dependences problem can be effectively eliminated after accounting for the second-order lag of the spatially-weighted average unemployment rates. The empirical results show that although regional characteristics still have certain explanatory power in affecting regional unemployment, the spatial interactions between regions are confirmed to be the key factors in Taiwan’s regional unemployment rates. The spatial spillovers decayed with distance as the regional unemployment disparities declined. Therefore, it can be confirmed that there exists a regional unemployment clustering effect in Taiwan. In addition, from the empirical results in this dissertation, the urban regions exhibit large positive spatial spillovers from their neighboring regions, that is, from their surrounding counties. Additionally, urban regions have higher unemployment rates and are more significantly affected by business cycles not by human capital. Moreover, when the surrounding county and its business center (corresponding major city) are re-classified as the same regional labor market, these regional labor markets show more spatial clustering on the geographical distribution. Furthermore, it is important to note that this is the first time in the literature that the possible influences to the variance of regional unemployment rates are discussed. An AR model is utilized to explore the possible influences of the variance of regional unemployment rates in Taiwan from 1988 to 2005. Although human capital does not affect regional unemployment rates significantly as we mentioned before, nevertheless, improvement in labor quality may be the key factor to reduce regional economic inequality in Taiwan’s regional labor markets. These dispersions between regions consistently have positive affects on the variance of regional unemployment rates, especially the deviation of the manufacturing sector between regions. It is worthwhile to mention that through empirical estimation, improvements in the highway transportation network may actually worsen the distortion between regions. |
