“已实现”跳跃检验与跳跃风险测度

发布时间：2018-01-15 00:10

本文关键词：“已实现”跳跃检验与跳跃风险测度　出处：《华中科技大学》2013年博士论文　论文类型：学位论文

【摘要】：进入21世纪以来,由于信息技术的快速发展,获取日内交易数据变得越来越容易,利用高频数据研究资产收益率的日内特征成为金融领域的一个新的热点话题。为了使资产收益率的建模既不违背市场无套利假定,又在数学上容易处理,一般假定收益率服从某个半鞅过程。学者们利用日内高频数据,采用非参数方法估计潜在波动,研究发现,已实现波动与已实现极差波动都是积分波动的无偏、一致的估计。在低频环境中,市场微结构噪声可以忽略不计,但在高频环境下,由于买卖价差、非连续交易、最小报价单位等微结构因素的影响,使得已实现波动一致高估积分波动,因此,“降噪”方法的研究成为金融计量研究的热点话题。除了微结构噪声外,资产价格跳跃也会导致已实现波动一致高估积分波动。因此,学者们构造了许多已实现估计量,例如二幂次变差和拉普拉斯已实现波动,既对跳跃稳健,又是积分波动无偏、一致的估计。为了甄别资产价格中跳跃成分,学者们提出了许多跳检验统计量,有些跳检验对微结构噪声很稳健,例如ABD检验和LM检验,有些跳检验的检验功效很高,例如CPR检验和PZ检验。本文沿用CPR检验的思想,利用已实现极差估计,构造新的跳检验,并启发性地给出了它的大样本性质。有些资产价格跳跃只受本公司或者本行业消息的影响(定义为异质跳跃),而有些资产价格跳跃只受整个市场消息的影响(定义为系统性跳跃)。依据资产组合理论,只受本公司或者本行业消息影响的异质跳跃风险可以被一个足够大的资产组合所分散,而那些系统性跳跃风险则是不可分散的。如果资产价格跳跃存在不可分散的成分,那么现有的资产定价与风险管理理论将受到巨大挑战。A股市场存在系统性跳跃吗?这是一个值得研究的问题。本文分别利用指数-个股法和mcp方法检验A股市场的系统性跳跃,研究结果表明,A股市场的系统性跳跃是显著存在的,且两种检验方法的检验结果差异很小。本文通过理论推导证明了指数-个股法的严谨性,通过引入阈值改进了等权二幂次变差的小样本性质。本文将系统性跳跃和异质跳跃视为极端事件,从极值理论的视角探讨股票收益率分布的尾部特征,利用TOD方法消除高频数据的日内效应,运用指数-个股法分解系统性跳跃和异质跳跃,并采用POT方法分别估计它们的左尾和右尾参数。实证研究表明,A股市场日内效应具有明显的“L”型特征,每支股票的系统性跳跃与异质跳跃都是显著存在的,且两类跳跃都具有非常明显的厚尾特征,所有股票的右尾跳跃次数和贡献都大于左尾。这表明,频繁出现的资产价格跳跃及其尾部特征是导致股票收益率非正态分布的一个重要原因。为了从系统性跳跃风险这一微观层面探讨贝塔系数的时变特征,本文利用“已实现”方法分解连续性贝塔和跳跃性贝塔,并分别检验连续性贝塔和跳跃性贝塔的稳定性。研究结果表明,短期连续性贝塔稳定性较差,中期和长期连续性贝塔比较稳定,而短期、中期和长期跳跃性贝塔的稳定性都很差。因此,短期贝塔系数的不稳定主要来自于连续性贝塔,而中期和长期贝塔系数的不稳定则来自于跳跃性贝塔。资产价格跳跃不仅是系统性的,还可能是自激励的。本文在新的]3AR-CJ-M模型框架下研究了沪深300指数隔夜风险的动态特征、影响因素以及可预测性,利用BN-S方法将日内波动分解为连续性波动和跳跃性波动,并运用ACH模型估计发生跳跃的意外性程度,进而采用最小二乘和分位数回归方法估计日内波动率指标和跳跃的意外性程度对隔夜风险的影响。研究结果表明,日内连续性波动、跳跃性波动和隔夜风险的滞后项都会显著地影响隔夜风险,且存在不对称效应；日内跳跃对大的隔夜风险的影响非常显著,且可以利用HAR-CJ-M模型很好地预测大的隔夜风险。这表明,日内跳跃会向前传导至隔夜跳跃,跳跃的自激式影响是显著存在的。
[Abstract]:Since twenty-first Century, due to the rapid development of information technology, access to intraday data becomes more and more easy, with characteristics of high frequency data on asset return days has become a new hot topic in the financial field. In order to make the model of asset returns without violating the market no arbitrage assumption, and easy to handle in Mathematics in general, it is assumed that yield obey a semi martingale process. Scholars using high-frequency intraday data, study on potential volatility estimation using non parametric methods, realized volatility and realized range volatility fluctuations are integral unbiased, consistent estimates.
In the low frequency environment, market microstructure noise is negligible, but in high frequency environment, due to the sale of price, non continuous trading, the effects of tick size and other micro structure factors, the realized volatility consistent overestimates therefore, "integral fluctuation, noise reduction method research has become a hot topic of research. In addition to financial measurement the micro structure of noise, asset price jumps will lead to overestimate the realized volatility consistent integral fluctuation. Therefore, scholars have constructed many realized estimators, such as the two power variation and Laplasse realized volatility, which is robust to jump, integral fluctuation unbiased, consistent estimates. In order to jump component screening assets the price, scholars have put forward many jump test statistic, some jump test on the microstructure noise is very robust, for example, ABD test and LM test, some jump test the effect of the test is very high, such as CPR test and PZ In this paper, we use the idea of CPR test to construct a new jump test by using the estimated maximum difference, and illuminate its large sample properties.
Some asset prices jump effect only by the company or the industry news (defined as heterogeneity, and some asset price jumps) effects of jump only by the whole market news (defined as the systematic jump). On the basis of portfolio theory, the heterogeneity only by this company or the industry news jump risk can be a sufficient the portfolio is dispersed, and the systematic jump risk is undiversifiable. If the asset price jumps there cannot be dispersed components, then the existing asset pricing and risk management theory will be a huge challenge in.A stock market there is a systematic jump? This is a problem worthy of study in this paper. The system of stock index - jump method and MCP method to test the A stock market, the results show that the system of A stock market jump is significant, and the two kinds of test methods test results difference It is very small. This paper proves the rigor of the exponential - share method by theoretical deduction and improves the small sample property of the equal weight two power difference by introducing a threshold.
The system of jumping and jumping as heterogeneous extreme events, explore the characteristics of tail stock returns distribution from the perspective of extreme value theory, the elimination of the high frequency data of intra day effect by using TOD method, decomposition system jump and heterogeneous jump using index - a stock method, and uses the POT method of their left and right tail tail parameter estimated respectively. The empirical research shows that, A stock market intraday effect has obvious characteristics of "L", each stock systematic jump and heterogeneous jumps are significant, and have obvious fat tail characteristics of two kinds of jump, all stock right tail jumping times and contribution are greater than the left tail.. this suggests that the frequent asset price jumps and the tail is the cause of stock returns characteristics of non normal distribution is an important reason for systematic jump risk. From the micro level of beta time varyingcharacteristics, The realized method of decomposition of continuity and jumping beta beta, were tested the continuity and stability of beta beta jump. The results show that the short-term continuous beta stability is poor, medium and long term continuous beta is relatively stable, while the short-term, medium-term and long-term stability of the beta jump are very poor. Therefore, short-term beta instability mainly from the continuity of the beta, while the medium and long term beta instability from jumping beta.
Asset price jump is not only systematic, but also may be self excitation. The dynamic characteristics of the]3AR-CJ-M model in the framework of new research on the Shanghai and Shenzhen 300 index overnight risk, influence factors and predictability, using the BN-S method to intraday volatility into continuous volatility and jump volatility, and the use of ACH model the estimated jump accident, and then using the least squares and quantile regression method to estimate the impact of intraday volatility index and the degree of jumping accident risk overnight. The results showed that days of continuous volatility and jump volatility risk and overnight lag will significantly affect overnight risk, and asymmetric effect; days jump effect on overnight risk is very significant, and the HAR-CJ-M model can be used to well predict the overnight risk. This suggests that the days ahead will transfer to the historic jump The night jump, the self excited effect of jumping is significant.

【学位授予单位】：华中科技大学
【学位级别】：博士
【学位授予年份】：2013
【分类号】：F224;F830.91

【相似文献】