Nonparametric Jump Test Method Based on Nonzero Drift Process and Its Effectiveness Research
The occurrence of jumps can cause a huge impact on the financial market,and the traditional BN-S jump test method assumes that the drift in the jump-diffusion model is zero.However,the phenomenon of asset bubbles in the financial market in the short term can lead to the emergence of nonzero drift.This paper proposes a solution to the jump testing problem in the case of nonzero drift.Firstly,the nonzero drift term is eliminated by subtracting the intraday mean(or median)from the intraday logarithmic return.Then,the test statistic is reconstructed and its asymptotic consistency under the limit theory is illustrated.Finally,simulations are carried out to show the feasibility of the method as well as the validity of the new test statistics,and the difference of using mean or median to eliminate non-zero drift is compared.It is applied to the Shanghai Composite Index for empirical analysis,and then compared with the existing LM method considering nonzero drift.The results show that the method can better test the jump problem with nonzero drift under finite samples,and is equally effective without considering overnight returns.
NETL
NSTL
万方数据
