Tail and dependence behavior of levels that persist for a fixed period of time (Q2271707)

Let \(X_1\), \(X_2\),\dots be a stationary sequence and let \(Y_i=\min(X_i,\dots, X_{i+s})\) for some fixed \(s\). The authors investigate the tail behavior of \(Y\) in connection with the tail behavior of \(X\). Three cases are considered: i.i.d. \(X_i\); \(X_i\) being an ARMAX process: \(X_i=\max(cX_{i-1},Z_i)\) for some \(0<c<1\) and i.i.d. \(Z_i\); \(X_i\) being an ARMAXp process: \(X_i=\max((X_{i-1 })^c,Z_i)\). The tail index and extremal index of \(Y_i\) are calculated, Leadford and Tawn's coefficient of tail dependence \(\eta\) is obtained for \((Y_1,Y_{1+m})\).