Comparison of Estimators of Tail Dependence Coefficient

To find effective estimations of tail dependence, we present the estimators of upper tail dependence coefficient by using survival copula. We research to two problems by using the samples from t-copula. Firstly, do the estimators estimate effectively the upper tail dependence coefficients of copula? Which is the best among the estimators? Secondly, if sample isn't from true distribution, do the estimators have model risk? Simulation shows that the estimators can estimate effectively the tail dependence coefficients of t-copula. By comparison of estimators, a better estimator is found out among the estimators. Therefore if sample is from true distribution, the estimators can estimate effectively upper tail dependence coefficients. However, if sample isn't from underlying copula, the estimators have model risk.