Hence ηM (X) → 0 if a- (X) → 0, ∀ i. The first order condition for an
optimal portfolio of claims is
χ-γ = λ- ∙ φ(X); ∀(i,X).
Hence
— ■ --1 1
xi- = a-(X) = λ- γi [φ(X)]- γi/X,
X
so that ai(X)e(0, 1).
Differentiating ln(αi(X)) with respect to ln(X) yields
a- ( X )
a- ( X )
= ⅛1 ПМ ( X ) - 1]
X γi
(15)
As shown by Benninga and Mayshar (2000), η'M(X) < 0 and for X → ∞,
an → 1, so that ηM(X) → γn. Hence, for X → ∞, d-(X) → 0,∀i, and
ηM(x) → 0.
Now consider X → 0. Then η'M(X) → 0 if a-(X) → 0, i = 2, ..,n, since
ɪʌ αi(X) = 0. From the first order condition, optimal risk sharing implies
'∙ (⅛)- ■
Y 1
γ γi
x1
or
X-
X 2
a- ( X ) f λi ∖ γi xγi
X = U1) X2
As shown by Benninga and Mayshar (2000), α1(X) → 1forX → 0 so that
ηM (x) → γ1 and x1 → X. Hence the last equation yields for X → 0
αi ( X )
X
→ (λ- ) γi X ~γ1i 2 ,i = 2 ,..,n.
λ1
This term goes to zero for X → 0ifγ1 > 2γi,i =2, .., n. Then, by equation
(1), a- ( X ) → 0 for X → 0, i = 2 ,...,n. Hence η'M ( X ) → 0 for X → 0. ■
24