Author/Authors :
ARSLAN, Öğr. Gör. Sıddık Gazi Üniversitesi - Bankacılık ve Sigortacılık Yüksekokulu - Bankacılık Bölümü, Turkey , ÇELEBİOĞLU, Salih Gazi Üniversitesi - Fen-Edebiyat Fakültesi - İstatistik Bölümü, Turkey , ÖZTÜRK, Fikri Ankara Üniversitesi - Fen Fakültesi - İstatistik Bölümü, Turkey
Title Of Article :
STATISTICAL INFERENCES IN TWO DIMENSIONAL ARCHIMEDEAN COPULAS: A CASE STUDY
Abstract :
Copulas, has become widely used and useful tool in exploring dependences structure and modeling. Archimedean copula families, allow us to determine dependence structure based on the family parameters. In this study, the major characteristics of the bivariate Archimedean copulas has been explored and to forecast its family parameters, a nonparametric based method of Kendall’s τ has been utilized together with real data set application from financial institutions Furthermore, the method will be tested and results have been discussed. Also, USD and Euro daily Exchange rates data for the period of Feb. 03, 2007 – Oct.17, 2008 , total of 1489 data points has been used to model the relationship between these two variables. Further, non-parametric forecasting method together with its applications for the two dimensional Archimedean copulas has been applied. The study revealed that, although two dimensional Archimedean copulas have been proved to be an important forecasting methodology, but including Lambda function or K (v)’ in the process of forecasting has found to be strengthen the quality of forecasting. Besides, experimental Lambda function, at 95% confidence interval and theoretical lambda functions applied for the three Archimedean copula families has been drawn.
NaturalLanguageKeyword :
Copula , Archimedian copula , dependence structure , exchange rates , non , parametric forecasting.
JournalTitle :
Journal Of Economics and Administrative Sciences
