基金經理人與投資人間的代理問題起源於兩者目標的不一致，前者要求個人薪酬財富的極大化，後者要求投資組合價值的極大化。造成目標歧異的原因有兩個：其一是在資訊不對稱的環境下，投資人無法觀察到經理人投資組合是否真正從投資人利益角度出發，因此引發了經理人的道德危險；其二是在競爭激烈的基金產業□，年度定期績效評比結果與經理人薪酬紅利多寡相連結等制度的設計，加重經理人選擇持有一個高風險投資組合的逆誘因，特別是期中累積績效較差的輸家，隨著年終總績效結算日期的接近，在自利動機的驅使下可能會透過較高的操作風險調整幅度，企圖扭轉頹勢以求反敗為勝。 本研究首先利用卡方檢定、t檢定與Logistic迴歸模式，實證國內基金經理人是否具有自利性風險調整的行為傾向，接著從契約設計的觀點，以理論模式推導命題的方式，探討三種不同型態的誘因費契約對於抑制經理人自利性風險調整行為的作用，並間接利用問卷調查的方式來驗證其效果。研究的結果發現： 1.國內基金經理人不管是贏家或輸家，在越接近年終總績效的結算時，都會偏向選擇一個高風險的操作水準，同時，上述特性在非外資型投信公司所發行的基金、新基金、小規模基金、資淺經理人所操盤的基金特別明顯。 2.基金投資人對於季等短期績效的過分重視，是導致國內經理人操作風險調整幅度偏高的主要原因之一。 3.純粹誘因費契約以及只加上「上限條款」設計的契約，這兩種契約都無法抑制經理人冒高風險的傾向；而純粹誘因費契約加上「懲罰條款」的設計，有抑制經理人自利性調高操作風險行為的效果。 本研究成果的貢獻主要有兩點： 1.在政策應用上，本研究提出契約條款設計的理念，對於我國擬開放勞退等大型基金委託代客操作，雙方契約該如何設計以確保投資人的權益，有相當參考的價值，本研究同時也對投信公司、基金經理人等提出相關的制度性建議。 2.在理論推導上，本研究融合一般化均衡分析法與選擇權理論的應用，將不同型態的誘因費契約化成經理人向投資人所購買的歐式買權，標的資產為經理人所持投資組合的價值，履約價格為比較基準指數的價值，執行日期為一年期的績效評比，模型導証的重點是經理人所選擇投資組合的風險程度與該類比選擇權價值的關係。 / An important question for the contracting literature is the extent to which real behavior is driven by the financial incentives contained in compensation schemes. To address this issue, (1) we use the tournament concept as the framework, and focus on the competitive nature of mutual fund environments how to affect the managers' portfolio decision-making processes. (2) we also use the Black-Scholes option pricing model as the framework, and analyze the impact on the mutual fund managers' risk-manipulation behavior of a performance-based incentive plan. Given the asymmetric information financial markets, most investors of mutual funds can not measure the funds' risk without error, thus, agent-divergent behavior may potentially arise. In a tournament reward structure, the managers' rational attempting to maximize their expected compensation may revise the risk level or alter the composition of their portfolio during the assessment period. While there will be times when such changes don't serve the best interest of funds' investors. Our research demonstrates the following results: 1. An empirical investigation of 86 open-type mutual funds during 1995 to 1998 with the methods of test, t test and the Logistic regression shows fund managers with poor performance would become aggressive and tend to increase fund volatility in the latter part of an annual assessment period. The effect is obviously clear toward the end of the year and it is involved with the investors' myopic of the assessment to the managers' performances. 2. In our model, the bonus is similar to a call option on the funds' portfolio. Three types of incentive contracts are compared. The results show that the incentive contract with penalty can reduce managers to adversely alter the risk of the portfolio they manage. It dominates the pure incentive contract and only with a ceiling incentive contract. Questionnaires investigated by fund managers will support some hypothesis.