49
R⅛n∖t) = exp{-⅛⅛n∖t)}
= ew-V________
n i⅜∑U"ΛH<>)
∑li<1∕≈.,ι⅛"-ltfe)
-V_____-_________}
⅛⅛∑Z=ι Qlm-1∖¾)e≈p(w'∕3^) j
∑≈i≤^MΓ^1∖¾)
= exp{-V-------ʒ-------,—,-, -,∖ } (3.20)
Note that the proposed Sθm∖t) and Rθm∖t) are also copula-based, which are dif-
ferent from Sθ0∖t) and /f,ɑŋ(/.).
Similarly as Step 1, we then update corresponding F^m∖β, G^l∖∙), P^m∖β, Qffl∖∙β
⅛n>(∙) and £<”>(■).
Step 3,
Again, maximize lβrββ.βc') and get the updated ^(m+1) and βiβn+lV
Step 4,
Iterates as below:
• Keep updating dfm+1) and β,βn+^ using the Newton-Raphson Method until they
converge.
• Following values are used to get new estimates of β and βc∙.
β(-m+1β βcm+1∖ Pi'n∖∙'), Qi"i∖∙), D<fl∖β and Ê-m\-). That is, after getting new