Advanced Microeconomics - Problem set 1

Due date: classes on Novemeber 6

Consider the following utility function, called CES (constant elasticity of substitution func-
tion):
u(x1 , x2 ) = (xρ1 + xρ2 )1/ρ , where 0 6= ρ < 1.
This function is obviously strictly increasing.
Problem 1 (5p). For the above CES function:
1. Show that u is strictly quasiconcave.
Solution. Take two bundles x, x0 ∈ R2+ , such that u(x) = u(x0 ) = u. Define x00 =
tx + (1 − t)x0 for t ∈ (0, 1). We must show that u(x00 ) > u. We have:
u(x00 )ρ = (tx1 + (1 − t)x01 )ρ + (tx2 + (1 − t)x02 )ρ
> txρ1 + (1 − t)x0ρ ρ 0ρ
1 + tx2 + (1 − t)x2
= tu(x)ρ + (1 − t)u(x0 )ρ = uρ
where the inequality follows from the fact that function y ρ is strictly concave for 0 6= ρ < 1.
So by strict monotonicity of the power function we obtain: u(x00 ) > u as claimed.

2. Formulate the Utility Maximization Problem and solve it to derive Walrasian demand
functions x1 (p, w) and x2 (p, w) [You may assume interior solution and Walras law which
follows from strict monotonicity of u(·)].
Solution. The UMP is as follows:
maxx1 ,x2 ≥0 (xρ1 + xρ2 )1/ρ
s.t. p1 x1 + p2 x2 ≤ w
The Lagrangian for this problem is L(x1 , x2 , λ) ≡ (xρ1 + xρ2 )1/ρ + λ(w − p1 x1 − p2 x2 ). The
FOCs assuming interior solution and Walras law are:
∂L(x1 , x2 , λ)
= (xρ1 + xρ2 )1/ρ−1 xρ−1
i − λpi = 0, for i = 1, 2
∂xi
∂L(x1 , x2 , λ)
= w − p1 x1 − p2 x2 = 0
∂λ
Solving for x = (x1 , x2 ) gives:
1
pi1−ρ w
xi = xi (p, w) = ρ ρ , for i = 1, 2.
p11−ρ + p21−ρ
ρ
Or after defining r = ρ−1 :

pr−1
i w
xi (p, w) = , for i = 1, 2.
p1 + pr2
r

3. Form the indirect utility function v(p, w).

Solution.
ρ  r−1 ρ #1/ρ
"
pr−1
1 w p2 w
v(p, w) = u(x(p, w)) = r r +
p1 + p2 pr1 + pr2
p1 + pr2 1/ρ
 r 
= w
(pr1 + pr2 )ρ
= w (pr1 + pr2 )−1/r
 4. Show that the indirect utility function is:

(a) homogeneous of degree zero in (p, w),

Solution. For any λ > 0:

v(λp, λw) = λw ((λp1 )r + (λp2 )r )−1/r

= λλ−1 w (pr1 + pr2 )−1/r
= w (pr1 + pr2 )−1/r
= v(p, w)

(b) increasing in w and decreasing in p.

Solution.
∂v(p, w)
= (pr1 + pr2 )−1/r > 0
∂w
∂v(p, w)
= −w (pr1 + pr2 )−1/r−1 pr−1
i < 0, for i = 1, 2.
∂pi

5. Verify the Roy’s identity.

Solution.

∂v(p, w)/∂pi −w (pr1 + pr2 )−1/r−1 pr−1

i
− = −
∂v(p, w)/∂w (pr1 + pr2 )−1/r
pr−1
i w
=
p1 + pr2
r

= xi (p, w), for i = 1, 2.

Problem 2 (5p). For the above CES function:

1. Formulate the Expenditure Minimization Problem and solve it to derive Hicksian demand
functions h1 (p, u) amd h2 (p, u) [You may assume interior solution and no excess utility
which follows from continuity of u(·)]

Solution. The EMP is as follows:

minx1 ,x2 ≥0 p1 x1 + p2 x2
s.t. (xρ1 + xρ2 )1/ρ ≥ u

The Lagrangian for this problem is L(x1 , x2 , λ) ≡ p1 x1 + p2 x2 + λ u − (xρ1 + xρ2 )1/ρ . The

FOCs assuming interior solution and no excess utility are:

∂L(x1 , x2 , λ)
= pi − λ(xρ1 + xρ2 )1/ρ−1 xρ−1
1 for i = 1, 2
∂xi
∂L(x1 , x2 , λ)
= u − (xρ1 + xρ2 )1/ρ = 0
∂λ
Solving for x = (x1 , x2 ) gives:
1
xi = hi (p, u) = u(pr1 + pr2 ) r −1 pr−1
i , for i = 1, 2.

2. Form the expenditure function e(p, u).

2
 Solution.

e(p, u) = p1 h1 (p, u) + p2 h2 (p, u)

1 1
= u(pr1 + pr2 ) r −1 pr1 + u(pr1 + pr2 ) r −1 pri
1
= u(pr1 + pr2 )(pr1 + pr2 ) r −1
= u(pr1 + pr2 )1/r

3. Show that the expenditure function is concave in p.

Solution. Let’s take p00 = tp + (1 − t)p0 , where p, p0 >> 0. We must show that e(p00 , u) ≥
te(p, u) + (1 − t)e(p0 , u)

e(p00 , u)r = u[(tp1 + (1 − t)p01 )r + (tp2 + (1 − t)p02 )r ]

≥ u[tpr1 + (1 − t)p0r r 0r
1 + tp2 + (1 − t)p2 ]
= tu(pr1 + pr2 ) + (1 − t)u(p0r 0r
1 + p2 )
= te(p, u) + (1 − t)e(p0 , u)

where the inequality follows from the concavity of xr , where r < 1 (note that when 0 6= ρ <
ρ
1, then 0 6= r = ρ−1 <1)

4. Verify the Shephard’s lemma.

Solution.
∂e(p, u)
= r/ru(pr1 + pr2 )1/r−1 pr−1
i = hi (p, u)
∂pi

5. Show that the following relationships are true:

(a) v(p, e(p, u)) = u

(b) e(p, v(p, w)) = w
(c) h(p, v(p, w)) = x(p, w)
(d) x(p, e(p, u)) = h(p, u)
Solution. This is rather straightforward, plugging into obtained definitions.

6. Write Slutsky equation for the above functions.

Solution. We have the following:

∂hi (p, u)
= (pi pj )r−1 (1 − r) (pr1 + pr2 )1/r−2 u (1)
∂pj

∂xi (p, w) ∂xi (p, w)

+ = (pi pj )r−1 (1 − r) (pr1 + pr2 )−2 w (2)
∂pj ∂wxj (p, w)
Since we know that w = e(p, u) = (pr1 + pr2 )1/r u. Hence the above two formulas are the
same.