银行间同业拆借利率走势的统计分析

发布时间：2018-03-08 03:07

本文选题：同业拆借利率　切入点：影响因素　出处：《山东大学》2014年硕士论文　论文类型：学位论文

【摘要】：银行间同业拆借利率是金融机构尤其是银行间短期资金借贷的利率,利率水平受短期资金供求的影响,其变动影响着商业银行的收益性和安全性。上海银行间同业拆借利率自成立以来逐渐成为我国的基准利率,是我国利率市场化的重要步骤和体现。然而2013年“钱荒”引发的隔夜拆借利率猛涨至13.44%的事件引起了各方对资金流动性和同业拆借利率的关注。因此本文试图通过分析影响同业拆借利率的因素和建立合适的模型,来拟合其走势,为商业银行等金融机构规避利率风险提供一定的依据。本文使用上海银行间同业拆借利率(Shibor)作为同业拆借利率的数据来源,选取隔夜和一周同业拆借利率作为主要研究对象,选用其两者自2007年以来至2013年的所有报价,假定服从广义误差分布和t分布,探究最为合理的拟合模型。首先本文回顾了Shibor的研究文献,国内对Shibor的研究主要从影响因素、走势拟合和利率期限结构以及其他一些方面展开,其中利率期限结构近年来引起越来越多的关注。其次本文结合前期文献研究,从经济学角度分析了可能对同业拆借利率产生影响的因素,以及影响的方向和程度。第三部分论文介绍了在对隔夜和一周同业拆借利率研究中使用的主要理论和模型概念。再者本文使用ARMA和ARCH族模型对隔夜和一周Shibor进行分析拟合,寻求能最好地刻画其走势的模型,并且对各模型的优劣进行探讨；其中在广义误差分布和t分布假设下分别使用GARCH、EGARCH和TARCH模型拟合隔夜和一周同业拆借利率的走势在相关文献中较少出现。经探究发现,Shibor具有显著的自相关性和异方差性,ARCH族模型能更好的拟合隔夜和一周同业拆借利率的走势,其中具有非对称效应项的EGARCH和]GARCH模型更优。GARCH模型中ARCH和GARCH项系数之和大于1,即隔夜和一周Shibor对冲击的反应有扩大的持续的效应。EGARCH模型显示Shibor的变动在利率上升时比利率下降更大,说明同业拆借利率对好消息和坏消息的反应是非对称的。TARCH中隔夜和一周同业拆借利率的TARCH项系数为负,即Shibor具有反杠杆效应。这与之前文献对同业拆借利率的研究基本吻合。除Shibor隔夜序列EGARCH模型中更适用t分布外,其他情况广义误差分布都优于或近似于t分布假设下的模型拟合。
[Abstract]:The interbank lending rate is the rate at which financial institutions, especially banks, borrow short-term funds. The level of interest rates is affected by the supply and demand of short-term funds. The Shanghai Interbank offered rate has gradually become the benchmark interest rate of our country since its establishment. It is an important step and embodiment of the marketization of interest rate in our country. However, the overnight borrowing rate skyrocketed to 13.44% in 2013, which aroused the attention of all parties to the liquidity of funds and the interbank offered rate. After analyzing the factors that affect the interbank offered rate and establishing the appropriate model, To fit its trend, for commercial banks and other financial institutions to avoid interest rate risk to provide a certain basis. In this paper, the Shanghai Interbank offered rate (SIBOR) is used as the data source of the interbank offered rate (IBOR). The overnight and one-week interbank offered rates are selected as the main research object, and all the quotations from 2007 to 2013 are selected. Assuming the generalized error distribution and t distribution, the most reasonable fitting model is explored. Firstly, this paper reviews the research literature of Shibor. The domestic research on Shibor is mainly based on the influencing factors. Trend fitting, term structure of interest rate and other aspects are carried out, among which interest rate term structure has attracted more and more attention in recent years. From the angle of economics, this paper analyzes the factors that may influence the interbank offered rate. The third part introduces the main theories and model concepts used in the study of overnight and one-week interbank offered rates. Furthermore, this paper uses ARMA and ARCH family models to analyze and fit overnight and one-week Shibor. To find the best way to describe the trend of the model, and to explore the merits and demerits of each model; Under the assumption of generalized error distribution and t-distribution, the trend of overnight and one-week interbank offered rate fitting using GARCH-EGARCH and TARCH models respectively is rare in relevant literature. It is found that Shibor has significant autocorrelation and heteroscedasticity. The arch family model can better fit the trend of overnight and one-week interbank offered rate. The sum of the coefficients of ARCH and GARCH in the model of EGARCH and] GARCH with asymmetric effect is greater than 1, that is, the Shibor response of overnight and one week to shock has an extended sustained effect .EGARCH model shows that the change of Shibor is when the interest rate rises. More than interest rates, It shows that the reaction of interbank offered rate to good news and bad news is asymmetric. The TARCH coefficient of overnight and one week interbank offered rate is negative in .TARCH. That is, Shibor has anti-leverage effect, which is basically consistent with previous studies on interbank offered rate. Except for the Shibor overnight sequence EGARCH model, it is more suitable for t distribution. In other cases, the generalized error distribution is superior to or similar to the model fitting under the assumption of t distribution.
【学位授予单位】：山东大学
【学位级别】：硕士
【学位授予年份】：2014
【分类号】：F832.33;F224

【参考文献】