本文選題：金融風(fēng)險(xiǎn) + 在險(xiǎn)價(jià)值��； 參考：《上海交通大學(xué)》2012年碩士論文

【摘要】：風(fēng)險(xiǎn)的確切定義是當(dāng)資產(chǎn)價(jià)值、股票價(jià)格和盈利情況出現(xiàn)非預(yù)期結(jié)果時(shí)的不確定性。如何有效預(yù)測(cè)并度量風(fēng)險(xiǎn),對(duì)于風(fēng)險(xiǎn)管理有決定性意義。 對(duì)于目前股票二級(jí)市場(chǎng)重要的投資主體之一基金公司來說,也需要建立一套完善的風(fēng)險(xiǎn)預(yù)測(cè)和管理機(jī)制,投資者更為關(guān)注的是投資左半?yún)^(qū)間可能出現(xiàn)的最大虧損,即一定概率下可能的最大投資損失。通過調(diào)研文章發(fā)現(xiàn)目前H基金公司對(duì)于基金在股票市場(chǎng)風(fēng)險(xiǎn)管理和損失預(yù)測(cè)的量化方法主要是固定損失比例和近似VaR歷史模擬法兩種,并希望引入不同VaR方法進(jìn)行方法比較,從而選取出更為有效的預(yù)測(cè)和管理方法。本篇論文由此選取H基金旗下的兩種最典型基金類型作為研究對(duì)象,分別是：指數(shù)型股票基金和非指數(shù)型開放式股票基金。加入幾種不同的VaR方法(Delta-正態(tài)VaR,歷史模擬VaR, CVaR,方差-協(xié)方差VaR)和基金原有管理方法進(jìn)行比較。通過運(yùn)用Eviews和Excel工具并使用LR似然比后測(cè)檢驗(yàn)驗(yàn)證各種方法的準(zhǔn)確程度,從而得出其中比較有效的方法。 通過實(shí)證研究,可以得到的結(jié)論是： 一、對(duì)于指數(shù)型股票基金,分別進(jìn)行了一年期,三年期VaR預(yù)測(cè),并選取了不同的時(shí)間區(qū)間來屏蔽牛熊市帶來的趨勢(shì)影響,外加不同置信水平(99%、95%、90%)共12種情況。使用似然比后測(cè)檢驗(yàn)表明,歷史模擬VaR和Delta-正態(tài)VaR分別通過了的11次(91.7%)和10次(83.3%),優(yōu)于其他模型。CVaR方法在置信水平比較低時(shí)誤差較大,總共通過4次(33.3%)檢驗(yàn),而固定損失比例方法在大部分檢驗(yàn)中都無法通過,僅通過3次(25%)。最后用Jarque-Bera檢驗(yàn)驗(yàn)證了模型選取的指數(shù)是趨向于正態(tài)分布的,并用ARCH檢驗(yàn)驗(yàn)證指數(shù)不存在條件異方差情況,說明以上VaR的計(jì)算都是有效的。 二、對(duì)于非指數(shù)型開放式股票基金,方差-協(xié)方差VaR和歷史模擬VaR預(yù)測(cè)優(yōu)于其他模型,兩種方法通過了所有似然比后測(cè)檢驗(yàn),但歷史模擬方法的精確度更高,檢驗(yàn)結(jié)果更接近于各個(gè)置信水平的理論推薦值。同時(shí)發(fā)現(xiàn)固定損失比例預(yù)測(cè)在大部分檢驗(yàn)中無法通過。 最后文章建議H基金管理公司在今后的投資工作中加入Delta-正態(tài)模型和方差-協(xié)方差模型兩種VaR方法并保留歷史模擬VaR方法來對(duì)旗下基金進(jìn)行風(fēng)險(xiǎn)度量和管理。而固定損失比例的方法和CVaR由于不夠精確和有效建議不再使用。
[Abstract]:The exact definition of risk is uncertainty when the value of assets, stock prices, and earnings are unexpected. How to effectively predict and measure risk is of decisive significance to risk management. For fund companies, one of the most important investors in the secondary stock market, it is also necessary to establish a set of sound risk forecasting and management mechanisms. Investors are more concerned about the maximum losses that may occur in the left half of the stock market. That is, the maximum possible investment loss under a certain probability. It is found that the quantitative methods for risk management and loss prediction of H fund companies in stock market are mainly fixed loss ratio and approximate VaR historical simulation, and it is hoped that different VaR methods will be introduced to compare them. In order to select a more effective prediction and management methods. In this paper, two typical types of H fund are selected as the research object: index stock fund and non exponential open equity fund. Several different VaR methods such as Delta-normal VaR, historical simulation VaR, Cvar, variance-covariance VaR) are added to compare with the original fund management methods. By using Eviews and Excel tools and using LR likelihood ratio test to verify the accuracy of various methods, the more effective methods are obtained. Through empirical research, we can draw the following conclusions: Firstly, for the index stock funds, the one-year and three-year VaR forecasts are carried out respectively, and different time intervals are selected to shield the trend effects of the bull bear market. There are 12 kinds of cases in addition to the different confidence levels 99% 95% and 90%. Using the likelihood ratio post-test, it is shown that the historical simulation VaR and Delta-normal VaR passed 11 times 91.7) and 10 times 83.3% respectively, which is better than other models. The Cvar method has a big error when the confidence level is low, and it has passed 4 times of 33. 3) test. However, the fixed loss ratio method can not be passed in most tests, only three times. Finally, the Jarque-Bera test is used to verify that the index selected by the model tends to normal distribution, and the ARCH test is used to verify that the index does not have conditional heteroscedasticity, which shows that the calculation of the above VaR is effective. Second, for non-exponential open-end stock funds, variance-covariance VaR and historical simulation VaR prediction are superior to other models. The two methods have passed all likelihood ratio post-test tests, but the historical simulation methods have higher accuracy. The test results are closer to the theoretical recommended values of each confidence level. At the same time, it is found that the prediction of fixed loss ratio can not be passed in most tests. Finally, the paper suggests that H fund management company should add Delta-normal model and variance-covariance model to the future investment work and retain the historical simulated VaR method to measure and manage the risk of its funds. The fixed loss ratio method and CVaR are no longer used due to their inaccuracy and effectiveness.
【學(xué)位授予單位】：上海交通大學(xué)
【學(xué)位級(jí)別】：碩士
【學(xué)位授予年份】：2012
【分類號(hào)】：F224;F832.51

【參考文獻(xiàn)】

相關(guān)期刊論文 前5條

1 宋錦智;VAR值的三種估算方法及其比較[J];城市金融論壇;2000年12期

2 周澤炯;;基于GARCH模型的VaR方法對(duì)我國(guó)開放式基金風(fēng)險(xiǎn)的分析[J];經(jīng)濟(jì)管理;2006年22期

3 劉子斐;史敬;;VaR模型比較技術(shù)及其評(píng)價(jià)——理論、實(shí)證回顧及其應(yīng)用初探[J];金融研究;2008年05期

4 龔銳,陳仲常,楊棟銳;GARCH族模型計(jì)算中國(guó)股市在險(xiǎn)價(jià)值(VaR)風(fēng)險(xiǎn)的比較研究與評(píng)述[J];數(shù)量經(jīng)濟(jì)技術(shù)經(jīng)濟(jì)研究;2005年07期

5 姚京,李仲飛;基于VaR的金融資產(chǎn)配置模型[J];中國(guó)管理科學(xué);2004年01期

，

本文編號(hào)：1836821