GARCH预测的货币指数动态组合分析

发布时间：2018-08-10 22:34

【摘要】：本文首先由汇率价格和美元指数的表现形式引出货币指数和隐性基准货币的概念,作为在外汇市场上对货币建立投资组合的资产选择和估值基准。进而,本文主要探讨了两个问题,首先在动态时间尺度上,以历史数据通过选择合适的数据窗口长度和调整频率这两个参数建立多期组合,并得出权重变化依托的合适两个参数范围值；第二个是在第一个问题的基础上对货币指数投资组合引入预测因素时,有没有对组合模型效果进行改进的可能,进而本文对外汇市场上存在的八种主要货币进行了实证。本文在绪论部分简单介绍了投资组合理论在证券市场尤其在外汇市场的应用,并且介绍了货币指数和隐性基准货币是如何引入的以及它能成为货币投资组合基准货币的可能和必然性。然后回顾了现代投资组合理论自建立到如今近60年的发展历程及研究现状并重点综述了多期投资组合模型和最优投资消费模型。第三章首先应用马科维茨的经典均值-方差理论在单期条件下建立了八种货币指数的投资组合模型。然后引入两个外生变量移动窗口的宽度L和组合权重调整频率T,对多期条件下的组合权重调整进行了分析,探讨了一般情况下这两个变量存在合适范围值的可能性。在此基础上,针对均值-方差模型对输入数据可靠性的要求,本文试图引入预测的因素来替代用简单平均值来表示的期望收益率,因此引入了ARCH族模型预测的功能。在介绍完ARCH族模型的由来后,本文对八种货币指数的ARCH效应进行了检验,表明八种货币指数的波动均存在明显的条件异方差性,进而本文发现应用GARCH(1,1)模型可以解决这一问题,能够更好地拟合货币指数的历史价格数据。从而在此基础上,本文把GARCH模型的预测能力引入货币指数投资组合中,利用货币指数的预测对数价格数据进一步得到其预测收益率,以其替代均值表示的期望收益率。在此基础上加入预测所得新的样本值重新计算组合风险,从而重新构建货币指数的投资组合并利用历史数据进行了实证,结果发现货币指数的预测收益率组合模型相对于一般均值-方差模型的优势和不足之处,在实践上预测收益率组合模型有其适用的范围。本文最后在对全文进行总结后,对货币指数多期投资组合模型以及在加入预测因素后做了一些扩展性的讨论。
[Abstract]:This paper first introduces the concepts of currency index and implicit benchmark currency from the expression of exchange rate price and dollar index as the asset selection and valuation benchmark for establishing a portfolio of currencies in the foreign exchange market. Then, this paper mainly discusses two problems. Firstly, on the dynamic time scale, the historical data is used to set up the multi-period combination by selecting the appropriate data window length and adjusting frequency. And get the appropriate two parameter range value of weight change depending on; the second is on the basis of the first problem, when introducing the forecast factor to the currency index portfolio, there is no possibility to improve the effect of the portfolio model. Furthermore, this paper makes an empirical study of the eight major currencies in the foreign exchange market. In the introduction part, this paper briefly introduces the application of portfolio theory in the securities market, especially in the foreign exchange market. It also introduces how the monetary index and implicit benchmark currency are introduced and the possibility and inevitability that it can become the currency portfolio benchmark currency. Then it reviews the development and research status of modern portfolio theory from its establishment to the recent 60 years, and summarizes the multi-period portfolio model and the optimal investment consumption model. In the third chapter, Markowitz's classical mean-variance theory is used to establish the portfolio model of eight monetary indices under the condition of single period. Then, the width L of the moving window of two exogenous variables and the adjustment frequency T of the combined weight are introduced to analyze the combined weight adjustment under multi-period conditions, and the possibility of the existence of suitable range values for these two variables in general is discussed. On this basis, aiming at the reliability requirement of the mean variance model for input data, this paper attempts to introduce the prediction factor instead of the expected return rate expressed by the simple average value, so the function of the ARCH family model prediction is introduced. After introducing the origin of the ARCH family model, the ARCH effect of the eight monetary indices is tested in this paper. The results show that the fluctuation of the eight monetary indices has obvious conditional heteroscedasticity. Furthermore, it is found that the application of the GARCH (1K1) model can solve this problem. Can better fit the historical price data of monetary index. On this basis, this paper introduces the forecasting ability of the GARCH model into the monetary index portfolio, and uses the predicted logarithmic price data of the monetary index to further obtain its predicted return rate, which replaces the expected return rate expressed by the mean value. On this basis, the portfolio risk is recalculated by adding the predicted new sample value, and then the portfolio of monetary index is re-constructed and the historical data are used to make an empirical study. The results show that the predictive return portfolio model of monetary index has its advantages and disadvantages compared with the general mean-variance model. In practice, the forecasting yield combination model has its applicable scope. In the end, after summarizing the whole paper, the paper discusses the multi-period portfolio model of currency index and the extended discussion after adding the forecasting factors.
【学位授予单位】：电子科技大学
【学位级别】：硕士
【学位授予年份】：2012
【分类号】：F820;F224

【参考文献】