本文根據財政部財稅中心營利事業所得稅申報檔2004 年資料，以Battese and Coelli (1995)超越對數函數型態之隨機生產邊界模型，估計台灣7,953 家紡織業的生產 效率，並以產業群聚、廠商規模作為無效率迴歸式之估計函數。本文研究發現，無效 率迴歸式中之產業群聚結果為顯著的負相關，即廠商群聚確能提升個別廠商之生產效 率；中小紡織業與生產無效率、產業群聚與廠商規模之交叉項均呈現顯著的負相關， 表示中小企業比大企業更具有生產效率，且產業群聚對於中小企業而言，更能夠有效 的提升其生產效率，進而強化競爭實力。因此，政府應加強產業群聚之相關政策，以協助中小企業發展。

This study gathers the data information of the 7,953 firms in textile industry based on business tax data compiled by the Tax Data Center, Ministry of Finance in 2004. We adopt Battese and Coelli’s (1995) model of stochastic frontier function to evaluate the production inefficiency and inefficiency model simultaneously. The empirical results reveal that the SMEs are more efficient than large enterprises. Secondly, industry cluster indeed promotes the production efficiency of SMEs. Therefore, the government should formulate the industry cluster strategies as soon as possible.