Determining the optimal window length of the time-varying copula parameter: a simulation study
Penulis/Author
ATINA AHDIKA (1); Prof. Dr.rer.nat. Dedi Rosadi, S.Si.,, M.Sc. (2); Dr. Adhitya Ronnie Effendie, S.Si., M.Si., M.Sc. (3); Prof. Dr. Drs. Gunardi, M.Si. (4)
Tanggal/Date
24 2020
Kata Kunci/Keyword
Abstrak/Abstract
A time-varying copula is a multivariate cumulative distribution function that can accommodate the temporal dependence of random variables. To capture the temporal dependence of random variables using a copula, we can form a time variation on copula parameters. Some studies reported that a function of a latent variable following an autoregressive process defines time-varying copula parameters. The process consists of an autoregressive coefficient describing the influence of the parameters in previous times and a forcing variable representing the relationship between the marginal variables which utilize previous information over a certain number of window lengths of the observations. Previous studies reported that ARMA(1,10) defines time-varying copula parameters. However, the optimal parameters of time-varying copulas do not always require ten last observations. Therefore, we determine the optimal window length of the time-varying copula parameter by conducting a numerical simulation on four types of copula with different strengths and directions of dependence. The window length is defined as a new parameter and the optimal parameters can be obtained by finding the values producing the smallest AIC. We also provide a numerical simulation on how to improve the accuracy of the predicted series by changing the mean and standard deviation of the residual pairs.
Level
Internasional
Status
Dokumen Karya
No
Judul
Tipe Dokumen
Aksi
1
24_Determining the optimal window length of the timevarying copula parameter a simulation study.pdf
