建構策略性生產能力之估計方法(I)

Abstract

計畫名稱：建構策略性生產能力之估計方法 半導體和TFT-LCD 面板等高科技產業，對台灣經濟貢獻極大，是目前台灣經濟成 長的主要動力，但其面對的是高度的全球化競爭。在高科技產業的經營策略中，決定生 產設施的產能投資規模是一項極重要的議題，需能滿足未來市場需求，也牽動著市場競 爭力的消長。產能決策牽涉到各種不同複雜因素和極為龐大的投資金額，是一策略性的 長期投資，設施建置後的使用期限相當長，且缺乏變動彈性，會嚴重影響公司財務表現 以至於生存能力。其決策的品質相當重要，卻主要根基於資訊的品質。本研究希望建構 一決策輔助工具，特別著重在提供單一生產單位(設施)的可能生產能力之有效資訊，包 含投入資源和可獲得產出的估計，以彌補制定產能投資決策時客觀資訊不足的困境。 本研究的主要目的在瞭解和估計理想狀況下的生產能力，即任一系統的生產極限， 特別是在規模上的變化型態，以獲得設施產能特性資訊，幫助企業做長期策略產能投資 規劃決策。藉由過去經驗和同業相關資料，可以生產經濟學中的確定性(deterministic)生 產前緣模型(frontier models)為學理基礎加以推估。然而此方法論假設樣本資料具凸性 (convexity)，因此估計結果可能無法反應常見的形態，例如經濟規模；且傳統相關方法 並未考量資料的隨機抽樣誤差，因此本研究將分三大主題加以探討，分別為「大樣本情 境下之生產能力前緣型態分析」，「小樣本情境下之生產能力前緣型態分析」和「生產能 力前緣型態資訊整合與呈現」。研究成果可以對生產能力規模型態分析進行統計推論， 同時，更將所得的結果加以視覺化（visualize），使得由受測資料所構成的前緣可利用簡 單圖形表達。本研究所建構的策略性生產能力之估計方法除了產業實務有所貢獻之外， 在學理上更是新的突破，能做為樣本資料凸性假設之統計檢定。

Title: Estimating strategic production capacity High-tech industry, such as semiconductor and TFT-LCD panel manufacturing, significantly contributes to Taiwan's economy nowadays. How to enhance the competitive strength of high-tech industry becomes a key for Taiwan’s economy growth. Determining proper facility scale is one important issue in today’s competitive business environment. It is a long term strategic decision that provides production capacity to meet the future demand. The production facility capacity decisions involve with huge amount of investment and many different intangible factors. Once the decision is made, there is little flexibility to adjust, and any mistake will damage the financial performance of the firm. The decisions thus not only have impact on profits and market share, but also the sustainability of the firm. The decision quality depends on the quality of supporting information. Therefore, the ultimate goal of the project is to develop a decision aid for strategic production capacity investment by providing proper objective capacity information, particularly the scale performance. Production capacity is the ideal condition of a production process (or called production system), which is the limit or the frontier of its production capability. The project aims to estimate the production capacity, particularly focusing on the impact of the system scale. With data from the past experience or the industry peer, one can estimate the production capacity based on deterministic frontier models, which is a technique in the context of production economics. However, deterministic frontier models do not consider the stochastic behaviors and more importantly, assume convexity of the underlying production process. Some real cases such as economic scale or increasing returns to scale will this be incorrectly estimated by the models. Three issues, (1) production capacity estimation with infinite sample, (2) production capacity estimation with finite small sample, and (3) information integration and vitalization on production capacity, are studied. Technically speaking, the study will develop a statistical inference model to analyze the shape of the capability frontier including checking the convexity of the production frontier. There is limited literature related to these issues. If concluded successfully, a tool will provide more objective and suitable information for strategic capacity investment decision making; new knowledge will also contribute to academia.