產品多樣性、不確定性與製造績效：晶圓代工廠商之實地實證研究

Product Variety, Uncertainty and Manufacturing Performance: A Field Empirical Study of A Semiconductor Wafer Fabrication Company

本研究蒐集來自一家晶圓代工廠商的實地資料，實證檢視產品多樣性與不確定性對製造績效之影響，相較於既存之產品多樣性研究，本研究延伸探討：製造環境中的不確定性對時間、品質與生產力績效之影響，以及產品多樣性對製造績效之直接與間接影響。關於製造環境的不確定性，本研究依據作業研究之文獻設計三項變數衡量之：即需求變異性、製程時間變異性與到達時間變異性。實證分析結果顯示：產品多樣性與不確定性均對設備生產力與生產品質具有顯著的負向影響，對等候時間與生產週期時間具有顯著的正向影響，其中，產品多樣性不僅直接造成製造績效的降低，同時也會導致較高的需求變異性、製程時間變異性與到達時間變異性，進一步造成製造績效的下降。綜言之，本研究之實證結果指出：除了產品多樣性之外，不確定性亦為重要的成本動因，若欲達成有效的成本管理，應著重於降低製造環境中不確定性的影響；其次，產品多樣性會導致製造環境之不確定性的增加，因而對製造績效存在直接與間接影響，倘若忽略該間接影響可能低估產品多樣性之成本，進而導致次佳之產品多樣性決策；另一方面，本研究亦發現不同的產品多樣性衡量對製造績效的影響不同，顯示：針對不同的產品組合特性應採行不同的管理政策。

This study examines the impact of product variety and uncertainty on manufacturing performance using data from one semiconductor wafer fabrication company. We extend product variety studies by providing evidence on the performance impact of uncertainty, and the mechanisms through which product variety impacts uncertainty and performance. Three variability measures are used to capture the level of uncertainty in the manufacturing environment, which are demand variability, process time variability and arrival variability. The empirical analyses indicate that greater product variety not only has an adverse impact on equipment productivity, production yield, and time performance, but also leads to higher demand variability, process time variability and arrival variability, which cause further degradation of manufacturing performance. Besides, the empirical results also show that a greater level of uncertainty has an adverse impact on manufacturing performance. These findings have several management implications. First, uncertainty in the manufacturing environment is an important cost driver which firms should focus on to attain effective cost management. Second, the relevant cost of product variety may be underestimated if the indirect impact of product variety on performance is not considered. Third, different dimensions of product variety affect measures of manufacturing performance differently and thus should be managed in different ways.