【作者】 彭敏瑜；

【导师】 范小云；

【作者基本信息】 南开大学 ， 金融学， 2013， 博士

【摘要】 中国大陆经济发展快速,未来发展重点将由出口带动经济成长的引擎转为以消费及投资内需为主的双引擎；产业发展重点亦由制造业渐渐转型为金融、资本市场、其它服务业中心。为达成此目标,金融商品不断地推陈出新,尤其是衍生性商品,如期货、期权、人民币相关计价商品,REITs也是其中一项规划的产品。因此,研究美日REITs发展经验,探讨其投资组合风险,对中国而言有极为重要的现实意义。但以研究标的而言,大部分文献以单一市场为主,如美国、欧洲或日本的REITs为主,极少数进行地区性REITs动态关连性分析。就研究方法而论,大部分文献应用因子分析(Factor Analysis)、向量自我回归(VAR)、双变量GARCH波动率模型探讨相关议题,难以了解二变量间的静态及动态的相关程度。而就Copula相关性(dependence)、共移或外溢效果的研究文献仅少数几篇,显然其应用仍有较大的空间,且尚无文献应用在REITs的研究。所以,本文除了探讨美、日与台湾地区REITs的发展历史、优点及特色外,亦比较次贷风暴前后,美日两地REITs的相关性。在方法上,除了运用传统的相关分析外,更透过静态及动态Copula关联分析来进行比较。此外,本文亦应用历史模拟法,变异—共变异数法,极端值模型及动态Copula模型来求算美日REITs投资组合在次贷风暴前后的风险值及绩效,获致的具体实证结果如下：第一,以Normal Copula模型最佳,但相关程度极低。表明日本的REITs报酬率并不受到美国REITs报酬率的影响。但次贷风暴后,美日REITs报酬率的相关程度升高,但以Clayton Copula最佳。表明美国REITs报酬率会影响日本REITs报酬率。第二,本文分别采用动态Normal Copula及动态的Clayton Copula,结果发现次贷风暴后具有显着的传染效果。意涵美国REITs艮酬率变动会影响日本REITs报酬率。第三,正向及负向冲击的影响机率是有显着差异的。进一步言之,即当次贷风暴发生后,美国REITs对日本REITs的冲击影响明显变大。第四,比较不同的风险值模型,可以发现在次贷风暴前、后,历史模拟法及变异—共变异法所评估的风险值较极端值模型低,但Copula所评估的风险值平均又高于极端值模型；而经回溯测试结果可发现,大部分的风险值模型在95%、97.5%的显着水平下拒绝虚无假设,但总体以Copula表现最佳。本文的研究结果除了做为中国未来发展REITs的参考外,亦可做为全球REITs投资组合及风险管理的参考。更多还原

【Abstract】 Being the rapid economic development in China, it has been increased from the current development of important export-led economic growth engine converted to domestic consumption and investment-based twin-engine. What is more, its development of key industries increased from manufacturing, gradually transformed into the financial and capital market center. To achieve this goal, it continue to introduce new financial instruments, especially derivatives, such as futures, options, commodities denominated in RMB related, REITs are also one of the planning of the products. Therefore, it is very important in practical on the study of US and Japan’s REITs development experience and explores its portfolio risk.It has been a very long time in view of the development of global REITs. The experience of United States and Japan REITs development can make a reference. However, at my best knowledge about the topic, most of the literatures only discuss a single market, such as the U.S., Europe or Japan. REITs, and a little of literature discuss the regional REITs dynamically relationship.Regarding to the methodology, most of using Factor analysis, Vector Autoregression model (VAR), bivariate GARCH volatility model to discussed the related issues. It is difficult to understand two variables between the static and dynamic relationship. Just a few of the research literature applied the Copula correlation (dependence) in comovement or spillover effect, especially no any REITs article be used.Therefore, this dissertation discussed in addition to the U.S., Japan and Taiwan REITs development history, advantages and features outside. Also compares the subprime mortgage crisis before and during the US and Japan correlation of the two REITs. In empirical methodology, in addition to the traditional use of correlation analysis, but also through the static and dynamic Copula correlation analysis for comparison. In addition, this paper also uses historical simulation method, variance-covariance method, extreme value models and dynamic Copula model for calculating the Value at Risk(VaR) and performance of US and Japan REITs portfolio in the subprime mortgage crisis. The empirical results show that,1.Normal Copula model is the best, but the degree of correlation is very low which indicated that Japan’s REITs returns are not subject to the impact of U.S. REITs returns. But during the subprime mortgage crisis, U.S. and Japanese REITs returns have a higher degree of correlation, but Clayton Copula is the best. It implied that the United States will affect the Japanese REITs.2. Using static dynamic Normal Copula and Clayton Copula, found that the subprime crisis has significant contagion effects. Implications of changes in U.S. REITs returns will affect the Japanese REITs returns.3. The positive and negative shocks are the probability that there are significant differences.In other words, when the subprime crisis occurred, the U.S. REITs impact on the Japanese influence significantly larger.4. Comparing different Vary model can be found that in the sub-prime turmoil, after the historical simulation method and the variance-covariance method than the assessed Value at Risk model extreme values low, but the risk of Copula assessed value was higher than the average extreme value model. While back-testing results can be found through most of the Vary model at95%,97.5%significance level to reject the null hypothesis. The performance of Copula VaR model is still the best. In addition to the results of this study can take as a reference to China’s future development of REITs, it can also be used as global REITs portfolio and risk management reference.更多还原