1. Suppose ≿ is rational.

Show that
(a) ∼ satis es transitivity.
For any x, y, z ∈ X, if x ∼ y and y ∼ z, then we have x ≿ y, y ≿ x, y ≿ z, z ≿ y. As x ≿ y, y ≿ z,
we have x ≿ z. As z ≿ y, y ≿ x, we have z ≿ x. x ≿ z, x ≿ z ⇒ x ∼ z.
(b) ∼ satis es re exivity and symmetry.
For any x ∈ X, x ≿ x, x ≿ x ⇒ x ∼ x.
For any x, y ∈ X, if x ∼ y, then we can say x ≿ y, y ≿ x ⇒ y ∼ x.
(c) For any x, y, z ∈ X, if x ≻ y and y ≿ z, then x ≻ z.
x ≻ y ⇒ x ≿ y, x ≿ y, y ≿ z ⇒ x ≿ z
If x ∼ z, then z ≿ x; then z ≿ x, x ≿ y → z ≿ y; as y ≿ z, z ≿ y, then z ∼ y; then as ∼ satis es
transitivity, x ∼ y. It contradicts x ≻ y. So we have x ≁ z and x ≻ z.

2. Suppose Bob consumes only two goods, x1 and x 2, and has a lexicographic preference (see
lecture slides for de nition).
(a) Is Bob’s preference locally non-satiated? Show your arguments.
Yes.
∀x = (x1, x 2 ) ∈ X, ∀ϵ > 0, we can always nd 0 < ε < ϵ, and y = (y1, y2 ) = (x1 + ε, x 2 ) ∈ Bϵ(x)
and it is obvious that y ≻ x.
(b) Is Bob’s preference strictly monotonic? Show your arguments.
Yes.
x1 > y1
As per Lexicographic preference, x ≿ y ⇔ or .
x1 = y1 & x 2 ≥ y2
∀x, y ∈ X, x = (x1, x 2 ), y = (y1, y2 ), if x ≫ y, then x1 > y1, x 2 > y2 ⇒ x ≿ y. Also,
y1 < x1 and y2 < x 2 ⇒ y ⪰̸ x. x ≿ y, y ⪰̸ x ⇒ x ≻ y.

3. Let ≿ be a preference relation on X represented by u : X → ℝ. Show that, if u is

quasiconcave then the upper contour set of x ∈ X is convex. (i.e. ≿ (x) ≡ {y ∈ X | y ≿ x} is
convex.)
u is quasiconcave ⇒ ∀y, z ∈ X, ∀t ∈ [0,1], u(t y + (1 − t)z) ≥ min {u(y), u(z)}.
For any y, z ∈ ≿ (x), we have y ≿ x, z ≿ x,
quasiconcave
then u(y) ≥ u(x), u(z) ≥ u(x) u(t y + (1 − t)z) ≥ min {u(y), u(z)} ≥ u(x)
⇒ t y + (1 − t)z ≿ x ⇒ t y + (1 − t)z ∈ ≿ (x).

4. Draw two indifference curves of the following utility function. Make sure you label them
properly.
(a) u(x1, x 2 ) = min[2x1, x 2]
fi
fi
fl
fi
fi
fi
 (b) u(x1, x 2 ) = max[2x1, x 2]

(c) u(x1, x 2 ) = x1 + 3x 2

(d) u(x1, x 2 ) = min[2x1, x 2] + x1

5. Prove that if ≿ is strictly convex, then x ( p, I ) is a function. (i.e. UMP has a unique solution.)
If ≿ is strictly convex, we assume UMP has more than one solution. Without loss of generality, we
make x1, x 2 two of the solutions.

{u(x 2 ) ≿ u(x) ∀x ∈ {x ∈ X | p ⋅ x ≤ I}
u(x1) ≿ u(x) ∀x ∈ {x ∈ X | p ⋅ x ≤ I}
Then we have , make z = t x1 + (1 − t)x 2, t ∈ (0,1).

Obviously, p ⋅ z = t p ⋅ x1 + (1 − t)p ⋅ x 2 ≤ t ⋅ I + (1 − t) ⋅ I = I.

{u(x 2 ) ≿ u(x) ⇒ x 2 ≿ x
u(x1) ≿ u(x) ⇒ x1 ≿ x
, as per ≿ is strictly convex, z ≻ xi, i = 1,2, which means z ≿ x, ∀x ∈ X.

p ⋅ z ≤ I & z ≻ x, ∀x ∈ X contradicts the assumption that x1, x 2 are both the solution of UMP, so
UMP has a unique solution and x ( p, I ) is a function if ≿ is strictly convex.

6. Give the algebraic formulas of the Marshallian demand function of the following four types of
utility functions.
 (a) u(x1, x 2 ) = min [α1 x1, α2 x 2], α1, α2 ∈ N
B( p, I ) = {x ∈ X | p ⋅ x ≤ I}, the UMP: max u(x1, x 2 )
x∈B( p,I )
L = min [α1 x1, α2 x 2] + λ( p1 x1 + p2 x 2 − I )
∂L ∂u
∂x1
= ∂x1
+ λp1 = 0
∂L ∂u
∂x 2
= ∂x 2
+ λp2 = 0
∂L
∂λ
= p1 x1 + p2 x 2 − I = 0

{α2 x 2, if α1 x1 > α2 x 2 ∂x1

α1 x1, if α1 x1 ≤ α2 x 2 ∂u ∂u
As u(x1, x 2 ) = min [α1 x1, α2 x 2] = , = 0 or = 0.
∂x 2
So the system of equations above has no solution.
It is obvious that α1 x1 = α2 x 2 will always hold when utility is optimized. Otherwise, we can always
increase the utility by changing x1 & x 2.
Assuming that α1 x1 = α2 x 2 does not hold when utility is optimized. Without loss of generality,
assuming that α1 x1 > α2 x 2, then u = α2 x 2. Then we can consume less x1, x′1 = x1 − ϵ (ϵ > 0), and
p1 ⋅ ϵ
more x 2, x′2 = x 2 + . We can always nd ϵ > 0 to make α1 x1 > α2 x 2 still true. The new
p2
utility u′ = u(x′1, x′2 ) = α2 x′2 > α2 x 2 = u, so the assumption above is false and α1 x1 = α2 x 2 will
always hold when the utility is optimized.
I
x1( p, I ) = α
p1 + p 2 ⋅ α1
2
In summary, we can get I
x 2( p, I ) = α
p 2 + p1 ⋅ α2
1

(b) u(x1, x 2 ) = β1 x1 + β2 x 2, β1, β2 ∈ N

B( p, I ) = {x ∈ X | p ⋅ x ≤ I}, the UMP: max u(x1, x 2 )
x∈B( p,I )
L = β1 x1 + β2 x 2 + λ( p1 x1 + p2 x 2 − I ) + μ1 x1 + μ2 x 2
∂L ∂u
∂x1
= ∂x1
+ λp1 + μ1 = β1 + λp1 + μ1 = 0
∂L ∂u
∂x 2
= ∂x 2
+ λp2 + μ2 = β2 + λp2 + μ2 = 0
∂L
∂λ
= p1 x1 + p2 x 2 − I = 0
∂L
∂μi
= xi = 0, i = 1,2
As per KKT condition, we can have:
β2 I
i. If x1 = 0 & x 2 ≠ 0, we have μ1 ≠ 0 & μ2 = 0, then λ = − , x 2 = . We can also
p2 p2
p1 ⋅ β2 p β
nd that μ1 = − β1 > 0, then 1 > 1
p2 p2 β2
β1 I
ii. If x1 ≠ 0 & x 2 = 0, we have μ1 = 0 & μ2 ≠ 0, then λ = − , x1 = . We can also
p1 p1
p2 ⋅ β1 p β
nd that μ2 = − β2 > 0, then 1 > 1 .
p1 p2 β2






fi
fi
fi
 β1 β
iii. If x1 ≠ 0,x 2 ≠ 0, then μ1 = μ2 = 0, so λ = − = − 2 . In this situation, any
p1 p2
x ∈ B( p, I ) is the solution of UMP.
I p1 β1
x2 = p2
, when p2
> β2
I p1 β1
To conclude, x1 = p1
, when p2
< β2
p1 β1
∀x ∈ B( p, I ), when p2
= β2

(c) u(x1, x 2 ) = x1γ x21−γ ,

γ ∈ (0,1)
L = x1γ x21−γ + λ( p1 x1 + p2 x 2 − I )
∂L ∂u
= + λp1 = γ x1γ−1 x21−γ + λp1 = 0
∂x1 ∂x1 Equation 1
∂L ∂u Equation 2 γ ⋅ x2 p γ px
= + λp2 = (1 − γ)x1γ x2−γ + λp2 = 0 = 1 ⇒ = 1 1
∂x 2 ∂x 2 (1 − γ) ⋅ x1 p2 1−γ p2 x 2
∂L
∂λ
= p1 x1 + p2 x 2 − I = 0
λI (1 − λ)I
⇒p1 x1 = λ I, p2 x 2 = (1 − λ)I; x1 = , x2 = .
p1 p2
1
(d) u(x1, x 2 ) = (x1ρ + x2ρ) ρ , 0 ≠ ρ < 1
1
L = (x1ρ + x2ρ) ρ + λ( p1 x1 + p2 x 2 − I )
∂L ∂u ∂u
∂x1
= ∂x1
+ λp1 = ∂x1
+ λp1 = ρ1 (x1ρ + x2ρ) ρ−1 ⋅ ρx1ρ−1 + λp1 = 0
∂L ∂u ∂u 1
∂x 2
= ∂x 2
+ λp2 = ∂x 2
+ λp2 = ρ (x1ρ + x2ρ) ρ−1 ⋅ ρx2ρ−1 + λp2 = 0
∂L
= p1 x1 + p2 x 2 − I = 0
∂λ
x p x p 1 p 1
⟹ ( 1 ) ρ−1 = 1 ⟹ 1 = ( 1 ) ρ − 1 ⟹ x1 = ( 1 ) ρ − 1 ⋅ x 2
x2 p2 x2 p2 p2
p 1
⟹ p1 ⋅ ( 1 ) ρ − 1 ⋅ x 2 + p2 ⋅ x 2 = y
p2
1
y p2ρ − 1 ⋅ y
Solve for x 2: x 2 = 1
= ρ ρ
p1
p1 ⋅ ( p )ρ − 1 + p2 p1ρ − 1 + p2ρ − 1
2
1
y p2ρ − 1 ⋅ y
Solve for x1: x1 = 1
= ρ ρ
p2
p2 ⋅ ( p )ρ − 1 + p1 p1 ρ−1
+ p2 ρ−1
1

7. Given u( ⋅ ) is strictly increasing, continuous, and quasiconcave, show that x h( p, u) is a

convex set.
Assuming x1, x 2 ∈ x h( p, u), we have p ⋅ x1 = p ⋅ x 2 ≤ p ⋅ x, ∀x ∈ X and u(xi ) ≥ u, i = 1,2 as per
the de nition of Hicksian Demand.
Then let z = t x1 + (1 − t)x 2 (t ∈ (0,1)) ⇒ p ⋅ z = t ⋅ p ⋅ x1 + (1 − t)p ⋅ x 2 = p ⋅ x1 ≤ px, ∀x ∈ X.
fi
Also, as u( ⋅ ) is strictly increasing, continuous, and quasiconcave,
u(z) = u(t x1 + (1 − t)x 2 ) ≥ min [u(x1), u(x 2 )] ≥ u. So we can say z ∈ x h( p, u) and x h( p, u) is a
convex set.

8. Show that if u( ⋅ ) is homogeneous of degree one, then x h( p, u) and e( p, u) are homogeneous

of degree one in u.
u( ⋅ ) is homogeneous of degree one ⇒ u(t x) = t ⋅ u(x). We need to prove x h( p, t u) = t ⋅ x h( p, u)
and e( p, t u) = t ⋅ e( p, u).
x x
x h( p, t u) = arg min p ⋅ x s.t. u(x) ≥ t u ⇔ x h( p, t u) = arg min t ⋅ p ⋅ s.t. u( ) ≥ u
x∈X x∈X t t

⇔ x h( p, t u) = t ⋅ arg min
x
p ⋅ z s.t. u(z) ≥ u = t ⋅ x h( p, u).
z= t ∈X
h h
As per x ( p, t u) = t ⋅ x ( p, u) which is proved above,
e( p, t u) = p ⋅ x h( p, t u) = t ⋅ p ⋅ x h( p, u) = t ⋅ e( p, u).