HW4 Solution- Econ 500- Fall 2021

November 5, 2021

1
1.1 a
x3 can be derived using Walras’ law because the preference is non-satiated.

1.2 b
Well defined preference implies that
- x(p, w) is homogeneous of degree 0.
- x(p, w) satisfies Walras’ law.
- Slutsky substitution matrix s(p, w) is symmetric and negative semi-definite for
all (p, w).
First, κ = 1 because x1 is homogeneous of degree 0 .
Second, γ = 0 or σ = 1 because x2 is also homogeneous of degree 0 .
Note that
∂x1 ∂x1 5 δ
s11 = + x1 = − + x1
∂p1 ∂w p3 p3
∂x1 ∂x1 β δ
s12 = + x2 = + x2
∂p2 ∂w p3 p3
∂x2 ∂x2 β δ
s21 = + x1 = + x1
∂p1 ∂w p3 p3
∂x2 ∂x2 γσpσ−1
2 δ
s22 = + x2 = + x2
∂p2 ∂w p3 p3
(si3 and s3j are omitted. ) Third, The fact that s12 = s21 for all (p, w) implies
δ = 0 or x1 = x2 . Note that x1 = x2 holds when α = 100, β = γ = −5 and
σ = 1. Fourth, s11 s22 − s12 s21 ≥ 0 should hold for all (p, w).
2
- If γ = 0, then δ = 0 and s11 s22 − s12 s21 = − βp3 ≥ 0. Then β = 0
−5γ−β 2
- If δ = 0 and σ = 1, then s11 s22 − s12 s21 = p23
≥ 0 should hold. That is,
2
5γ ≤ β .
- If x1 = x2 , then s11 = s12 = s21 = s22 because β = γ = −5 and σ = 1.
Therefore,
(s11 s22 − s12 s21 ) = 0

1
Fifth, s11 ≤ 0 and s22 ≤ 0 for all (p, w). These conditions provide an restriction
on δ. For the detailed discussion, please see the remark. Some remark:
- You can calculate the components needed to get s.3 and s3 . using Walras’
law and the implicit function theorem. For example, we can calculate ∂x ∂w as
3

follows.
p1 x1 + p2 x2 + p3 x3 = w
implies
∂x1 ∂x2 ∂x3
p1 + p2 + p3 =1
∂w ∂w ∂w
∂x3 1 p1 ∂x1 p2 ∂x2 1 δ (p1 + p2 )
⇒ = − − = −
∂w p3 p3 ∂w p3 ∂w p3 p23
- You can check that s13 = s31 and s23 = s32 if δ = 0 or x1 = x2 even though
it is a tedious procedure.
- It is hard to restrict δ because x1 ≥ 0, x2 ≥ 0 and x3 ≥ 0 might set some
implicit restrictions on (p1 , p2 , p3 , w) . If s33 ≤ 0 is considered. The restriction
can be more complicated.
First, s11 ≤ 0 and s22 ≤ 0 can imply δ ≤ 0 because x1 ≥ 0 and x2 ≥ 0 Second,
when x1 = x2
∂x3 ∂x3
s33 = + x3
∂p3 ∂w
   
x3 p1 ∂x1 p2 ∂x2 1 p1 ∂x1 p2 ∂x2
= − − − + − − x3
p3 p3 ∂p3 p3 ∂p3 p3 p3 ∂w p3 ∂w
p1 x1 − 100 p2 x2 − α δ (p2 + p3 )
=− − − x3
p3 p3 p3 p3 p23
(p1 + p2 ) (x1 − 100 + δx3 )
=−
p23

This can imply δ ≥ 0 because s33 can be positive when δ < 0.

Therefore, we can guess δ = 0

2
MWG Proposition 3.H.1 is based on the following assumptions. - e(p, u) is con-
tinuous and differentiable.
- e(p, u) is strictly increasing in µ.
- e(p, u) is increasing in p.
- e(p, u) is concave in p.
- e(p, u) is homogeneous of degree 1 in p.

2
2.1 a
First z (p1 , p2 ) > 0 for all p because e(p, u) is strictly increasing in µ.
Second, z1 (p1 , p2 ) ≥ 0 and z2 (p1 , p2 ) ≥ 0 and α ≥ 0 because e(p, u) is increasing
in p. Third, z (p1 , p2 ) is concave and α ≤ 1 because e(p, u) is concave in p.
Fourth, the homogeneity in degree 1 of e(p, u) implies

z (λp1 , λp2 ) λα pα α
3 µ = λz (p1 , p2 ) p3 µ
⇒ z (λp1 , λp2 ) = λ1−α z (p1 , p2 )

That is, z (p1 , p2 ) is homogeneous of degree (1 − α).

Fifth, D2 e (p1 , p2 , p3 ) should be symmetric negative semi-definite. It requires
an additional restriction on z (p1 , p2 ) as well as the condition that z (p1 , p2 ) is
concave in p.

2.2 b
First, f (p1 + p2 + p3 ) should be homogeneous of degree 1 in p. This implies

f (λ [p1 + p2 + p3 ]) = λf (p1 + p2 + p3 )

Let’s pick λ = 1/ (p1 + p2 + p3 ). Then,

f (p1 + p2 + p3 ) = f (1) (p1 + p2 + p3 )

Then,
e(p, u) = f (1) (p1 + p2 + p3 ) µ
Because e(p, u) is strictly increasing in µ, f (1) > 0. Then, e(p, u) is concave in
p because p1 + p2 + p3 is a linear function. Also, you can see that e(p, u) is
increasing in p.
Then, substitution matrix is 0 so we cannot say that goods is net substitutes
or net complement. Note that all goods are gross complements to each other
because
w
xi (p, w) = h(p, v(p, w)) = f (1)v(p, w) =
p1 + p2 + p3
for all i.

3
e(p, v(p, w)) = w ⇒ ∂e(p,v)
∂w = 1 ⇒ ∂e(p,v) ∂u
∂u ∂w = 1
∂e(p,v) ∂e(p,v)
Envelope Theorem: ∂u = ∂µ and ∂w ∂u
= λ ⇒ ∂e(p,µ)
∂µ = λ1
The λ I’m talking about here is from UMP.
1
You can also solve this by the EMP and get λ from the EMP. λEM P = λU M P

3
4
Let’s denote zt as the bundle in year t. The table of expenditure is as follows.
z1 z2 z3 Revealed Preference
Year 1 36 36 30 z1 RD z2 and z1 RD z3

Year 2 43 41 36 z2 RD z3 and D

( not z2 R z1D

D
Year 3 45 39 38 ( not z3 R z1 and ( not z3 R z2
z1 RD z2 and z2 RD z3 imply z1 Rz3 . There is no violation to either WARP or
SARP. Therefore, the consumer’s behavior is consistent with utility maximiza-
tion theory.