Beruflich Dokumente
Kultur Dokumente
n n n
negative subset, R the strictly positive subset; R , R , R
second derivative f ' ' x = −1 x −2 (= the slope of f ' x at x).
concave iff:
semi-definite ∀ x X where
f 11 '' x ... f 1n ' ' x
∂2 f
H x = ... ... ... and f ij '' x = at x
∂ xi ∂ x j
f n1 '' x ... f nn ' ' x
det H x = f 11 ' ' x⋅f 22 ' ' x − f 12 ' ' x⋅f 21 ' ' x ≥ 0 , ∀ x X .
Strict concavity requires that the line joining any 2 points on the
(negative definite H x ).
For convex functions the line joining any 2 points on the graph lies
f 1 ' = x 2 , f 2 ' = x 1 : f 11 ' ' = f 22 ' ' = 0 , f 12 ' ' = f 21 ' ' = 1 ⇒ H x =
0 1
1 0
zero).
min]
1 1
f ' x = − 1 = 0 when x = 1 ; f ' ' x = − 2 0 ∀ x R ,
x x
So f is concave. So x = 1 is global max of f (& f 1 = 0 )
2 2
(b) f : R 2 R where f x1, x 2 = x 1 x2 − x1 − x 2
f 1 ' = 1 − 2x 1 , f 2 ' = 1 − 2x 2 ⇒ x =
1 1
,
2 2
is stationary point
f 11 ' ' = f 22 ' ' = −2 , f 12 '' = f 21 '' = 0 ⇒ H x =
−2 0 , n.s.d. so
0 −2
f is concave and x =
1 1
,
2 2
is global max
f
1 1
, =
2 2
1
2
.
CONSTRAINED OPTIMIZATION 1
max f x s. t. g x ≥ 0 (1)
x
(D) g x ≥ 0
1
Examples max x1 x 2 − x 12 − x22 s. t. 2x 1 2x 2 ≤
2
1
f is concave (previous exercise), g = − 2x 1 − 2x 2 is concave
2
(check) and g 0 0 (e.g.).
2 2
L = x1 x 2 − x 1 − x 2
1
2
− 2x 1 − 2x 2
∂L ∂L
(A) = 1 − 2x 1 − 2 = 0 = 1 − 2x 2 − 2 = 0 ,
∂ x1 ∂ x2
1 1
(B) ≥ 0 , (C) = 0 or 2x 1 2x 2 = , (D) 2x 1 2x 2 ≤
2 2
1 1
Try = 0 : x 1 = x 2 = from (A) but 2x 1 2x 2 = 2 no
2 2
1 1
solution. Try 2x 1 2x 2 = : x1 = x 2 = 1−2 from (A) so
2 2
1 3 1
21−2 = and = 0 : x1 = x2 = is solution
2 8 8
therefore.
is hod
2 12
(b) f x1 , x 2 = x 1 1 ⋅x 2 ⇒ f tx = tx 1 1 tx 2 2 = t ⋅f x and
f is hod 1 2
(c) f is hod ∑ i
(d) f is strictly concave if ∑ i 1 , concave if ∑ i = 1 , and
neither concave nor convex (if n ≥ 2 ) if ∑ i 1 .
The Cobb-Douglas as a utility function
2
max ux = x 1 1 x 2 s. t. p1 x 1 p2 x2 ≤ m x1 , x 2 ≥ 0
(u max) x ,
L x , = x1 1 x2 2 m − p1 x 1 − p2 x2
i −1
Since ui ' = i x i x j j 0 , u has no stationary points and the K-T
∂L 1−1 2 ∂L 1 2−1
(I) ∂ x = x
1 1
x 2
− p 1
= 0 = x
2 1
x 2
− p2 = 0
1
∂ x 2
1 x 2 p1 2 2
Dividing, ⋅ = ⇒ p x
2 2 = p x
1 1 ⇒ m = p x
1 1 p 1 x1
x1 2 p2 1 1
1 2
⇒ p1 x 1 = m , p2 x 2 = m
1 2 12
and the consumer's (Marshallian) demand functions are:
1 m 2 m
x1 p , m = ⋅ x2 p , m = ⋅
12 p1 12 p2
utility function;
1 2
1 m 2 m
[
V p , m = u x1 p ,m , x2 p , m = ] 1 2
⋅
p1 12
⋅
p2
Notice that, as prices and income vary, the budget share spent on
good i
pi x i
m
remains constant =
i
12 . Also the own price
i m pi
i = − ⋅ 2
⋅ = −1
12 pi i m
⋅
12 pi
These constant budget shares and unit elastic demands are special
expenditure function.
REFERENCES
244 – 250
Note: