半導體製造活動價值鏈中，半導體測試是重要的一環。半導體整合元件製造大廠為各種策略考量，於是漸提高向專業測試廠商尋求代工的比例，加上Design House產業蓬勃發展，也因此造就全球測試產業快速成長，甚至表現比整體半導體產業還優異。台灣半導體測試業為全球專業委外測試產業龍頭寶座。然而全球外在競爭環境威脅，而且在產業進入障礙低、競爭對手不斷增加、競爭日益強烈的情況下，終將直接衝擊台灣半導體產業在全球的地位。本研究欲從交易成本理論對測試委外服務品質和委外滿意度加以探討，以提供業界持續發展參考。 回顧過去委外相關研究發現，主要集中在：(1)企業是否應該委外或自製；(2)如何在統御結構下進行委外管理；(3) 採用討論或提出觀點說明的方式，少數真正採用實徵性研究驗證理論架構。因此，本研究專注於半導體測試委外的範疇，對委外進行探討，本研究架構乃為探討交易成本對服務品質和委外滿意度之影響，針對半導體測試產業型態及測試委外的需求建立研究架構與假設，並據以定義研究變數及制定問卷，經調查回收後做資料整理與統計分析。 經過資料分析，本研究在半導體測試委外產業的主要研究結果有三：一、在半導體測試外包產業中「交易成本」對於「服務品質」有顯著影響；二、「交易成本」對於「委外滿意度」未有顯著影響；三、「服務品質」不但對「委外滿意度」有正向的影響，同時亦扮演「交易成本」對於「委外滿意度」正向影響的「媒介變數」(mediator)。 根據上述的研究結果，本研究提出三點建議給予實務界參考。包括：一、以產能外包模式轉為整合模式外包，以降低不確定成本，提升服務品質；二、對於資產獨特性與不確定性質越高的委外案，承包商應提升對服務品質各方面的重視； 三、企業往後應慎重考量交易成本的影響。 / IC testing is a critical part of the semiconductor manufacturing value chain. In response to a variety of strategic considerations, large-scale IDM semiconductor manufacturers have increasingly sought the service from the professional testing firms. This, plus the rapid development of external IC design houses has created conditions for a global testing industry growing even more rapidly than the semiconductor industry as a whole. Taiwan’s specialized IC testing firms lead the global testing industry, but this dominance is increasingly threatened by external competition and low barriers to entry, and may eventually have a direct impact on the outlook of Taiwan’s IC industry. This study investigates transaction costs in terms of quality and customer satisfaction for external IC testing operations, and aims to provide the industry with a reference for sustainable development. Previous research in outsourcing has primarily focused on whether a company should outsource or keep operations in-house, or how to manage outsourcing within existing command structures, while a small number of empirical studies have used theoretical or case study methods to validate the use of theoretical frameworks. Therefore, this study focuses on the outsourcing of semiconductor testing operations to investigate how the structure and characteristics of transaction costs affect outsource service quality and satisfaction, with an aim to establish a research framework for investigating the semiconductor testing industry and outsourcing requirements. Towards this end, relevant variables have been defined and incorporated into a questionnaire which was used to source data for statistical analysis. Through statistical analysis, this study arrives at three primary conclusions regarding the outsourcing of semiconductor testing: (1) in the outsourcing of semiconductor testing, transaction costs have a significant impact on service quality; (2) transaction costs do not have a significant impact on outsourcing satisfaction; and (3) service quality is positively correlated with outsourcing satisfaction, and also plays transaction cost property in the positive correlation of outsourcing satisfaction and mediator. Based on these findings, this study raises three suggestions as a reference to practitioners: (1) move from a production outsourcing model to an integrated outsourcing model to reduce uncertainty costs and improve service quality; (2) for outsourcing cases with a higher degree of asset specialization and uncertainty, the contractor should focus especially on raising all aspects of service quality; and (3) enterprises should carefully consider the impact of transaction costs.

近年來全球學習華語文蔚為風潮，且隨著資通訊科技和網路快速發展，華語文教學更逐漸向「數位學習」發展。臺灣僑務委員會為提供海外華僑及外籍人士學習華語文，設立「全球華文網」，該網站至2011年為止已突破四千萬的瀏覽人次，使用者眾。本研究即以「全球華文網」為研究個案，旨在探討該網站的優使性，進而歸納網站的優勢和尚待改進之處。希望透過本研究了解華語文教學網站使用者平日需求，及理想的網站應具備哪些功能，對未來建置類似的華語文教學網站時有所助益。 本研究之研究目的可分為以下四項：(一)探討華語文教學網站的內容和網站使用者需求。(二)探討優使性評估定義與優使性測試方法。(三)探討華語文教學網站之優使性及使用者滿意度。(四)分析使用者優使性評估結果，找出網站缺失與改善方向，歸納建議作為未來建置華語教學網站之參考。 本研究採用深度訪談、優使性測試和優使性問卷調查三種研究方式，研究對象為華語文教學學位學程博碩士生及華語文教學學分學程大學生二者。透過實證研究得出以下五點結論：(一)華語文教學課程修習者常使用的網站可分為三類：資料查找、教學輔助、提供相關資訊；(二)使用者使用教學網站通常是任務導向；(三)使用者不滿意「全球華文網」之優使性；(四)「全球華文網」亟待加強網站之分類與名稱、資源整合、檢索功能三者；(五)使用者對「全球華文網」抱持正面且期待的態度。 根據以上結論，本研究也提出建議，希望作為建置類似網站時之參考。首先網站應從使用者的角度出發，思考如何給予最簡單有效的功能，如教學網站在建立前應先釐清建置目的和使用對象，才能進一步提供相對應之資料；網站的選單名稱應易於理解，各項資源應給予簡單說明；教學網站可就使用者之身分不同提供其相對應的使用介面和資源，以上各項功能用以幫助使用者節省摸索網站的時間。此外，網站也應定期進行使用者滿意度調查以及優使性評估，使用者提供的寶貴意見是網站改進的依據，而透過優使性測試，則能夠讓網站建置人員了解使用者操作的實際情形，彌平雙方的認知差距。定期的評鑑能讓網站了解優缺並即時改善和加強功能，以符合使用者的需求。

University of Macau / Faculty of Education

University of Macau / Faculty of Education

隨著數位相機、智慧型手機等影像記錄產品的普及，人們可以隨時隨地按下快門，使得個人相片的數量不斷的增長。倘若照片中缺少了註釋，進行搜尋檢索大量的個人照片是困難的。雖然如此，人們卻會因為繁瑣累人的照片註釋工作而缺乏意願。 本研究藉由Kuo所發展出的PARIS (Personal Archiving and Retrieving Image System)作為來源系統架構，其系統解決了大部分個人數位影像語義標註的困難，也利用半自動註記方法減少使用者註釋負擔。但由於過去缺乏行動載具，PARIS並未開發一個現代化的行動註釋系統，來協助一般使用者進行個人影像註釋。 為了解決此問題，本研究利用使用性工程進行系統的建構，以設計及開發出一套現代化的行動影像註釋系統iPARIS (Integrated Personal Archiving and Retrieving Image System)。iPARIS擴展了PARIS設計出符合行動載具的後設資料結構與本體，包含將PARIS所定義的DDDC後設資料標準擴充為得以群眾註記之社群媒體地理標籤的iDDDC (Integrated Dozen Dimensional Digital Content)，以及在既有的STO (Spatial and Temporal Based Ontology)的空間架構中加入更具彈性的地理標籤關聯方法GRO (Geotag Related Ontology)。iPARIS並利用工作流程設計、介面線稿設計、視覺設計、系統實作、使用者參與設計等步驟來設計及開發系統，再以迭代設計方法不斷地改善系統的使用者介面及使用者體驗。iPARIS在建置完成後利用參與啟發式評估量表、手勢次數、操作時間、操作路徑、錯誤次數進行使用性測試，而其測試結果也達預期的使用性目標。更進一步，透過分析使用者操作系統的實際行為來找出系統存在的使用性問題，並提出未來系統的改進方式。 透過完整的規劃與適切的設計系統介面，讓使用者在拍照、註釋、組織及管理照片都可以在行動應用程式上完成，並使此系統能滿足具有效率、具學習性、具記憶性、低錯誤率及高滿意度等Nielsen所提出的使用性標準，以降低個人影像註釋時的負擔，並增加使用者註釋個人照片的意願，以應付日益增長地個人數位影像資料的管理需求。 / With the widespread use of photograph recording products, such as digital camera and smartphone, people take photographs easily so that the number of personal photographs increases unceasingly. If photographs are lacking of annotation, searching and retrieving the large number of personal photographs become difficult. However, people are usually reluctant to the cumbersome and exhausted annotation work. In 2004, Kuo et al. developed PARIS (Personal Archiving and Retrieving Image System) to solve these problems. Based on MPEG-7, PARIS resolved most difficulties of personal digital image semantic annotation and lightened users’ annotation burdens by semi-automatic annotation methods. However, because of the lack of mobile device at that time, PARIS did not develop a modern mobile annotation system to help ordinary people annotate personal photographs. Therefore, this thesis builds a system with usability engineering as well as designs and develops a modern mobile image annotation system iPARIS (Integrated Personal Archiving and Retrieving Image System). iPARIS expands PARIS architecture to add the support for mobile. It expands PARIS DDDC metadata standard to iDDDC (Integrated Dozen Dimensional Digital Content) which is a crowdsourcing annotation enabled by third party social networking services, and provides STO’s spatial ontology a more flexible geo-information description methodology GRO (Geotag Related Ontology). iPARIS also takes the steps of workflow design, wireframe design, visual design, system implementation, participatory design to build the system; moreover, it keeps improving the user interface and user experience by iterative design. After implementing iPARIS, we conduct usability testing including participatory heuristic evaluation, gesture times, operation time, gesture path, error times, and interview. The test results reach the expected usability goals. Furthermore, the thesis finds some usability problems by analyzing users’ behaviors when they are using the application and brings future improvement for the system. By integral planning and appropriate interface design, this system lets user take, annotate, organize, and mange personal photos in a mobile application. We hope that iPARIS can meet the usability index to lighten the burdens of personal photograph annotation, arousing users’ desire to annotate personal photographs, and meet the growing needs for the management of personal digital photograph information.

根據AMR Research的預估，全球的客戶關係管理(Customer Relationship Management，CRM)市場規模，從1998年23憶美元，預估西元2003年將達到168憶美元。以及由2002年底Siebel推出的解決方案來看，CRM的趨勢將針對各種應用系統如何做好整合。 本研究之目的，旨在整合現有客戶關係管理相關文獻，與日商B公司多年來的經驗，以流程再造為出發點，從而設計出新的客戶管理資訊系統中之「銷售」、「行銷」以及「客戶服務」等機制，以解決舊系統不敷使用的問題，並且提供新商品線銷售與服務的機制，以及整合部份內外部系統。研究過程中實踐了系統開發生命週期(System Development Lifecycle)的理論進行新系統之分析、設計、開發、測試，並進一步分析與探討新系統實際上為B公司帶來的效益。以期對於國內企業因應快速發展的CRM趨勢，本系統的實現提供了國內其它產業進行CRM資訊系統建置過程的參考價值。 / According to the prediction of AMR research, the marketing scale of global Customer Relationship Management will be expected to rise from 2.3 billions in 1998 to 16.8 billions in 2003. Viewing the CRM solution of Siebel in the end of 2002, the trend of CRM will focus on how to integrate other hundreds of applications. The objective of this study is to integrate the related references of present CRM and the publisher industry experiences within a Japanese company B. On the basis of Business Process Reengineering (BPR) theory, we design the Sales, marketing, and customer services in the new CRM system in order to solve the problems on the legacy system, to provide the new sale channels and services mechanism of new product line and to integrate parts of the inner and 3rd party information system. During the research process, we accomplished the theorem of System Development Lifecycle to analyze, design, develop, and test of the new system. And further analyze and discuss of the benefit to company B from the the new system. In order to match the trend of the rapid growth in the CRM market for local enterprises, this system supplied the referenced value during building the CRM information system for other local enterprises. / 目 錄 摘要……..………………………………………………………… I 目錄………….…………………………………………………… II 表目錄……………………………………………………………. IV 圖目錄………………….………………………………………… V 第一章 緒論………………….…………………………………... 1 1.1. 研究背景與動機……….……………………………… 1 1.2. 研究目的與範圍………………………………………. 3 1.3. 研究方法與論文結構………………………………….. 4 第二章 文獻探討…………………………………………………6 2.1顧客關係管理系統………………………………………..……6 2.2企業資訊系統整合………………………………….…..…..…10 2.3企業策略委外…………………………………………….……12 2.4企業流程再造…………………………………………….........13 第三章 案例背景與解決方案………………………………..….15 3.1案例背景分析…………………………………………………16 3.1.1 舊系統功能……………………………………….…...17 3.1.2 舊系統業務流程………………………………………25 3.1.3 舊系統流程分析………………………………………27 3.2 解決方案……………………………………………………...29 3.2.1 新系統流程分析………………………………………30 3.2.2 新系統功能設計重點…………………………………31 3.2.3 預期效益………………………………………………33 3.3 技術需求分析………………………………………………...34 3.3.1 系統分析與設計方法…………………………………34 3.3.2 技術選用………………………………………………41 第四章 系統架構…………………………………..…………….44 4.1系統分析………………………………………………..……..44 4.1.1 業務流程圖……………………………………………45 4.1.2 Data Modeling………………………………………….50 4.2 系統設計……………………………………………………...57 4.2.1平行設計……………………………………………….58 4.2.2 MVC設計概念………………………………………...59 4.2.3 系統雛形……………………………………………....62 4.2.4其他系統整合介面…………………………………….68 4.2.5系統架構圖…………………………………………….69 4.3 系統開發………………………………………………….…..71 4.3.1 Web功能模組……………………………………….....72 4.3.2 批次模組………………………………………………74 4.4 系統測試……………………………………………………...76 4.4.1系統整合測試………………………………………….77 4.4.2使用者接受度測試…………………………………….81 4.5 系統上線……………………………………………………...82 4.5.1教育訓練……………………………………………….82 4.5.2上線計畫……………………………………………….82 4.6 系統開發相關技術…………………………………………...86 4.7 效益評估……………………………………………………...91 第五章 結論與後續研究方向…………………………………..98 5.1 結論…………………………………………………………..98 5.2後續研究方向…………………………….…………………..99 參考文獻………………………………….………………….101 表目錄 表2-1 四大巨人CRM解決方案的銷售與趨勢分析表…………………………...8 表3-1 銷售系統功能簡述………………………………………………………….17 表3-2訂戶資料管理功能簡述……………………………………………………..19 表3-3 契約(訂單)管理功能簡述…………………………………………………...20 表3-4 債權管理功能簡述………………………………………………………….22 表3-5催款管理功能簡述…………………………………………………………..22 表3-6發貨管理功能簡述…………………………………………………………..23 表3-7舊系統功能分析……………………………………………………………..24 表3-8新系統解決方案重點………………………………………………………..29 表3-9新系統各部門人員動作概要說明…………………………………………..30 表3-10流程分析策略分析表………………………………………………………36 表3-11軟體生命週期模式比較表…………………………………………………38 表 3-12 系統建構模式與軟體品質評析表………………………………………..42 表3-13 J2EE與 .Net比較表……………………………………………………….43 表4-1 Table Schema 範例…………………………………………………………..52 表 4-2 系統測試分類表……………………………………………………………76 表4-3 Unit Test測試案例…………………………………………………………...79 表4-4 SIT測試案例………………………………………………………………...80 表4-5 上線計畫表執行細項……………………………………………………….84 圖目錄 圖1-1研究方法……………………………………………………………………….5 圖3-1舊系統功能架構圖一(依子系統) ……………………………………………18 圖3-2舊系統功能架構圖二(依部門) ………………………………………………18 圖3-3 舊系統訂購/出貨流程……………………………………………………….26 圖3-4 新系統訂購/出貨流程圖…………………………………………………….31 圖3-5 ERD範例……………………………………………………………………..35 圖3-6 應用系統平行開發的方……………………………………………………..40 圖3-7 系統雛形設計流程圖………………………………………………………..41 圖4-1 B公司整體業務流程圖………………………………………………………45 圖4-2 客戶註冊/訂購流程………………………………………………………….46 圖4-3個人資料變更流程……………………………………………………………46 圖4-4 訂單內容變更………………………………………………………………..47 圖4-5 信用卡入金流程……………………………………………………………..47 圖4-6 退款流程……………………………………………………………………..48 圖4-7 客戶服務……………………………………………………………………..48 圖4-8 物流再發送流程……………………………………………………………..49 圖4-9 名條列印與線上分析報表流程……………………………………………..49 圖4-10 Conceptual Data Model (E-R with Entity) ………………………………….51 圖4-11 Conceptual Data Model Sample(E-R with attributes)……………………….52 圖4-12 亞洲共通版系統設計概念圖………………………………………………58 圖4-13 各子系統功能概要圖……………………………………………………....59 圖4-14 i18n的技術應用範例……………………………………………………….60 圖4-15第一層級Prototype範例(訂單) ……………………………………………63 圖4-16第二層級Prototype範例(訂單) …………………………………………..64 圖4.17第三層級Prototype範例(訂單) …………………………………………..65 圖4-18新系統架構圖……………………………………………………………..69 圖4-19系統硬體部署圖…………………………………………………………..70 圖4-20 JDBC Connection Pooling Module………………………………………..88

隨著全球電子商務的蓬勃發展，為在資訊流通與資料保護之間取得平衡，跨境資訊流動議題在國際經貿整合中往往也成為各國談判的焦點之一。其中，TPP中各國更進一步地處理「當地設立伺服器」議題，並提案將以「必要性測試」作為各國施行該措施之前提，本文遂就該測試在相關措施下之運作進行探討，並針對我國未來制度上與國際接軌之方向提出建議。 本文透過文獻分析法，歸納出必要性測試的評估過程中有「替代性措施對政策目標之貢獻程度」、「政策目標或價值之重要性」、「對國際貿易之限制性」幾個重要因素，另外在針對澳洲措施的分析中，有司法管轄權、技術、國際間合作狀況與行政成本，需納入考量之因素。然而，我國法制中尚未有更具體的管制措施，而經貿協議中僅止於承認彼此之資訊流通管制規範。未來或可參考澳洲之做法，同時由前歸納之因素評估該作法之必要性。 / The issue of trans-border data flow has been treated as one of those significant trade liberalization topics while global electronic commerce continues to surge and countries are striving to build common ground on the balance between data flow as well as data protection. Among these efforts of trade integration, “necessity test” was proposed when the members within TPP go further to negotiate whether the regulation of “localization data center” should be banned or not. In this article, we will conduct analysis toward how will the test work with potentially disputing measures and provide suggestions for Taiwan’s legislation to be geared to international treaties and standards. Through documentary analysis, factors evaluated against possible alternative measures when exercising necessity tests are summed up as contribution made by the compliance measure to the enforcement of the law or regulation at issue, the importance of the common interests or values protected by that law or regulation, and the accompanying impact of the law or regulation on imports or exports. Furthermore, as what has been shown by analyzing the Australian measure, we concluded that jurisdiction, techniques, international cooperation and administrative costs should also be taken into account. However, due to the lack of more practical measures and the topic in Taiwan’s concluded trade agreements is still in its infancy, I suggested that related authority can take the Australia’s measure as an example on basis of the above-mentioned factors.

投資決策時，除了關注資產的報酬外，更不可忽略風險。而風險的衡量上，一般常用風險值來衡量投資所面臨的風險，這是由於風險值具有動態管理、量化風險等優點。而國內研究對於不動產風險值的文獻上多以住宅市場為主，對於商用不動產較無著墨，是故本研究欲從不同風險值模型探討投資商用不動產的風險值，並分為直接投資(北市商用不動產)與間接投資(REITs)商用不動產兩個不同次市場。實證結果發現直接投資商用不動產風險值高於間接投資商用不動產。其次，本研究試圖比較靜態與動態風險值模型在估計不動產風險值的行為表現，經回溯測試(Back Testing)檢驗後發現，發現兩個模型衡量不動產風險值時，表現差異性不大。最後，本研究並以夏普績效(Sharp Ratio)來衡量直接投資與間接投資商用不動產的投資績效，研究期間為2007年6月至2009年3月。實證結果發現，直接投資商用不動產在景氣衰退與股市劇跌時具有抗跌性；而間接投資商用不動產則與股市發生同時下跌的現象，此現象可能是我國REITs具有代理問題(Agency Problem)與系統風險(Systematic Risk)等問題所致。因此，本研究建議投資者投資REITs時，應同時考量REITs存在上述的風險與問題，以避免投資上的損失。 / When making investment strategies, aside from considering the return of investment, one cannot ignore the risk factors. In measuring risk, we usually use VaR (Value at Risk) to calculate the risks of investment because, among other reasons, VaR has dynamic and quantifiable advantages. Most of the studies regarding real estate investment risk in Taiwan focus on residential markets; thus, this paper investigates commercial real estate markets using different VaR models to determine the degree of risk, distinguishing further between direct investment markets and indirect investment markets like REITs (Real Estate Investment Trust). The result of this study reveals that direct real estate investment involves higher risks than indirect real estate investment. Furthermore, there was hardly any difference in investment risk when using either static or dynamic VaR models in the computations after using Back Testing. Finally, this study employs Sharp Ratio to calculate commercial real estate investment performance covering the period between June 2007 and March 2009. Direct real estate investment shows firmness during economic downturns or stock market crashes unlike indirect real estate investment like REITs which follows stock market trends. This phenomenon may be due to Agency Problem and Systematic Risk in Taiwan’s REITs market. Therefore this study suggests that when investing in REITs one has to take into account the risks and problems in order to avoid unnecessary investment losses.

本論文為探索性的個案研究架構，從「知識管理活動」觀點來看工廠的知識管理作法。並經由「知識管理關鍵因子」與「知識管理活動」相互關連性來探討「知識管理關鍵因子」對於個案LED廠的「知識管理活動」之影響。 對於知識管理流程的運作得到的結論有： （一）、外部知識與內隱知識同為個案企業在知識取得及創造重要來源 但由個案探討發現要讓知識取得及創造能更有效果的方法，有賴激勵制度及高階主管全力支持以促使團隊合作發展才能達到目的。 （二）、由個案跨國企業知識蓄積及分享而言，IT及網際網路科技應用是必要且影響甚鉅的，但必須搭配資深或技術人員為載體，知識蓄積及分享效果才為顯著。 （三）、教育、評核及激勵制度推動師徒制及專案團隊與跨廠區、部門合作，促進知識流通、移轉及轉換。 對於知識管理關鍵因子與知識管理流程關聯性得到結論有： 1. IT及網際網路可協助知識管理活動更加有效率及組織化。 2. 科技產品的應用對遠距離、跨區域公司的知識管理式可收事半功倍之效果。 3. 信任與自私為一體兩面，沒有相對高階的絕對支持及相對資源投入，就無以產生絕對信任。知識導入及轉移至同化是否能達效果，此為一大主因。 4. 高階主管的認知及全力支持是知識管理落實、有效執行的關鍵。 5. 相同制度用在不同地區未必可得到一致效果，因地制宜因勢利導方可收知識管理之功效。 6. 使命與榮譽感需配合評核及有效激勵制度才能成為知識傳承重要動力。 / The thesis architecture is base on exploratory case study . Base on the point view of Knowledge management to discuss the activities of factories . Base on the inter-related “ key factors of knowledge management ” and “knowledge management activities ” . We try to find the impact from“ key factors of knowledge management ” to " knowledge management activities ”in the case study. The conclusion of the knowledge management process: 1. External knowledge and implicit knowledge are equal important resource for the case company in the knowledge acquisition and creation. However, the case study found that to make knowledge acquisition and the creation can be more effective is depends on the incentive system and top managers full support to promote the development of teamwork to achieve the purpose. 2. For knowledge accumulation and sharing of multinational enterprises , IT and Internet technology application is necessary and a great influence .It will be more powerful if senior technical staff as a carrier of knowledge accumulation and sharing . 3. Education、assessment and incentive system encourage apprenticeship and project teams to work more powerful in cross-plant .That promote the circulation of knowledge, transfer and conversion. We study the relationship between key factors of knowledge management and knowledge management process . Conclusion as following： 1. IT and Internet can make knowledge management activities more efficient and organized. 2. the application of technology products on the type of long-distance, inter-regional knowledge management can get maximum effect. 3. Trust and selfishness is two side of the same coin. If there is no absolute support and the relative resources from top managers . There will be no absolute trust . It’s the main key point of knowledge import 、transfer 、 assimilation . 4. Top managers awareness and fully support to knowledge management will be the key to effective implementation. 5. The same system in different regions may not be obtained consistent results. Capitalize on the trend will be the best efficacy of knowledge management. 6. Mission and sense of honor must comply with the assessment and effective incentive system in order to become an important driving force of knowledge transmission.

風險值大多是在假設資產報酬為常態分配下計算而得的，但是這個假設與實際的資產報酬分配不一致，因為很多研究者都發現實際的資產報酬分配都有厚尾的現象，也就是極端事件的發生機率遠比常態假設要來的高，因此利用常態假設來計算風險值對於真實損失的衡量不是很恰當。 針對這個問題，本論文以歷史模擬法、變異數-共變異數法、混合常態模型來模擬報酬率的分配，並依給定的信賴水準估算出風險值，其中混合常態模型的參數是利用準貝式最大概似估計法及EM演算法來估計；然後利用三種風險值的評量方法：回溯測試、前向測試與二項檢定，來評判三種估算風險值方法的優劣。 經由實證結果發現： 1.報酬率分配在左尾臨界機率1％有較明顯厚尾的現象。 2.利用混合常態分配來模擬報酬率分配會比另外兩種方法更能準確的捕捉到左尾臨界機率1％的厚尾。 3.混合常態模型的峰態係數值接近於真實報酬率分配的峰態係數值，因此我們可以確認混合常態模型可以捕捉高峰的現象。 關鍵字：風險值、厚尾、歷史模擬法、變異數-共變異教法、混合常態模型、準貝式最大概似估計法、EM演算法、回溯測試、前向測試、高峰 / Initially, Value at Risk (VaR) is calculated by assuming that the underline asset return is normal distribution, but this assumption sometimes does not consist with the actual distribution of asset return. Many researchers have found that the actual distribution of the underline asset return have Fat-Tail, extreme value events, character. So under normal distribution assumption, the VaR value is improper compared with the actual losses. The paper discuss three methods. Historical Simulated method - Variance-Covariance method and Mixture Normal .simulating those asset, return and VaR by given proper confidence level. About the Mixture Normal Distribution, we use both EM algorithm and Quasi-Bayesian MLE calculating its parameters. Finally, we use tree VaR testing methods, Back test、Forward tes and Binomial test -----comparing its VaR loss probability We find the following results: 1.Under 1% left-tail critical probability, asset return distribution has significant Fat-tail character. 2.Using Mixture Normal distribution we can catch more Fat-tail character precisely than the other two methods. 3.The kurtosis of Mixture Normal is close to the actual kurtosis, this means that the Mixture Normal distribution can catch the Leptokurtosis phenomenon. Key words: Value at Risk、VaR、Fat tail、Historical simulation method、 Variance-Covariance method、Mixture Normal distribution、Quasi-Bayesian MLE、EM algorithm、Back test、 Forward test、 Leptokurtosis