Then {1 - (vy, +)B +7328 Ae x, 
Vi +v2=—b], 
V1 ¥2= bz 
(1—v,B)(1—v9B) €, =X 
ee x 1 
~ B(B) (1+5,B+b2B2) (1—7,B)(1—v>B) 
€; Tt 
; 15 (vi Ute tobe 
VILRUA 
co yi vfs) 1 
= vat Se ees (5.188) 
FEO Vi -—V2 J 
Therefore, if the expression of the inverse function J; is used here as 
é:= >) CI) Xj, 
i=0 
Mina yitl . , : 
Tj=- a J vyi+ Z v3. (5.189) 
Ui vy) v2—-V¥V1 ilo?) 
This relation, as well as the way to derive it, is the same as that used for the derivation of 
Green’s function for AR(2), already shown in Section 5.2.3. For invertibility of MA(2), 
€,; must be bounded byt — t—j, j — ©. Then by the same logic as was used for AR(2), 
lvy1<1, lv21< 1 should stand for MA(2), and the conditions that must be satisfied by 
bj, bz are the same as the conditions that had to be satisfied by a), a2 of AR(2). This con- 
dition is shown in Fig. 5.31, which is similar to Figs. 5.15 and 5.16, where a), a2 are 
replaced by b,, b2. This triangular zone is again subdivided into subzones [I] and [I], 
depending on 
b? = 4b,[I] and b?< 4b,[II]. 
160 
