Vous êtes sur la page 1sur 27

Solutions to selected exercises from

Jehle and Reny (2001): Advanced


Microeconomic Theory
Thomas Herzfeld
September 2011
Contents
1 Mathematical Appendix 2
1.1 Chapter A1 . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 2
1.2 Chapter A2 . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 6
2 Consumer Theory 15
2.1 Preferences and Utility . . . . . . . . . . . . . . . . . . . . . . . . . . . . 15
2.2 The Consumers Problem . . . . . . . . . . . . . . . . . . . . . . . . . . . 17
2.3 Indirect Utility and Expenditure . . . . . . . . . . . . . . . . . . . . . . . 20
2.4 Properties of Consumer Demand . . . . . . . . . . . . . . . . . . . . . . 22
2.5 Equilibrium and Welfare . . . . . . . . . . . . . . . . . . . . . . . . . . . 24
Remark: Some answers might not be the most elegant ones from a mathematical per-
spective. Any comment and suggestion, also in case of obscurities, are highly welcome.
The graphs are not always perfect.
1
1 Mathematical Appendix
1 Mathematical Appendix
1.1 Chapter A1
A1.7 Graph each of the following sets. If the set is convex, give a proof. If it is not
convex, give a counterexample.
Answer
(a) (x, y)[y = e
x
This set is not convex.
Any combination of points would be outside the set. For example, (0, 1) and (1, e)
(x, y)[y = e
x
, but combination of the two vectors with t =
1
2
not:
1
2
(0, 1) +
1
2
(1, e) = (
1
2
,
e + 1
2
) / (x, y)[y = e
x
.
(b) (x, y)[y e
x
This set is convex.
Proof: Let (x
1
, y
1
), (x
2
, y
2
) S = (x, y)[y e
x
. Since y = e
x
is a continuous
function, it is sucient to show that (tx
1
+ (1 t)x
2
, ty
1
+ (1 t)y
2
) S for any
particular t (0, 1). Set t =
1
2
. Our task is to show that
_
1
2
(x
1
+ x
2
),
1
2
(y
1
+ y
2
)
_

S.
1
2
(y
1
+ y
2
)
1
2
(e
x
1
+ e
x
2
), since y
i
e
x
1
for i = 1, 2. Also,
1
2
(e
x
1
+ e
x
2
) e
1
2
(x
1
+x
2
= e
x
1
2
e
x
2
2
e
x
1
+ e
x
2
2e
x
1
2
e
x
2
2
e
x
1
2e
x
1
2
e
x
2
2
+ e
x
2
0 (e
x
1
e
x
2
)
2
0.
(c) (x, y)[y 2x x
2
; x > 0, y > 0
This set is not convex.
The graph shows that the function has a maximum at (0,0). To provide a formal
example,
_
1
10
,
1
2
_
,
_
1
9
10
,
1
2
_
S = (x, y)[y 2x x
2
; x > 0, y > 0. However,
_
1,
1
2
_
=
1
2
_
1
10
,
1
2
_
+
1
2
_
1
9
10
,
1
2
_
/ S
(d) (x, y)[xy 1; x > 0, y > 0
This set is convex.
Proof: Consider any (x
1
, y
1
), (x
2
, y
2
) S = (x, y)[xy 1; x > 0, y > 0. For any
t [0, 1],
(tx
1
+ (1 t)x
2
)(ty
1
+ (1 t)y
2
) = t
2
x
1
y
1
+ t(1 t)(x
1
y
2
+ x
2
y
1
) + (1 t)
2
x
2
y
2
> t
2
+ (1 t)
2
+ t(1 t)(x
1
y
2
+ x
2
y
1
), since x
i
y
i
> 1.
= 1 + 2t
2
2t + t(1 t)(x
1
y
2
+ x
2
y
1
)
= 1 + 2t(t 1) + t(1 t)(x
1
y
2
+ x
2
y
1
)
= 1 + t(1 t)(x
1
y
2
+ x
2
y
1
2) 1 i x
1
y
2
+ x
2
y
1
0.
2
1 Mathematical Appendix
x
1
y
2
+ x
2
y
1
= x
1
y
1
y
2
y
1
+ x
2
y
2
y
1
y
2
2
y
2
y
1
+
y
1
y
2
2 0
y 1 2y
1
y
2
+ y
2
0
(y
1
y
2
)
2
0,
which is always true and therefore, (tx
1
+ (1 t)x
2
, ty
1
+ (1 t)y
2
) S which is
convex.
(e) (x, y)[y ln(x)
This set is convex.
Proof. Let (x
1
, y
1
) + (x
2
, y
2
) S. Then
1
2
(y
1
+ y
2
) (ln(x
1
) + ln(x
2
)).
S is convex
if

1
2
_
ln(x
1
) + ln(x
2
) ln(
1
2
x
1
+
1
2
x
2
)
_

1
2
ln(x
1
x
2
) ln(
1
2
x
1
+
1
2
x
2
)
(x
1
x
2
)
1/2
(
1
2
x
1
+
1
2
x
2
)
x
1
2(x
1
x
2
)
1/2
+ x
2
0

_
x
1/2
1
+ x
1/2
2
_
2
0
which is always true.
6 4 2 0 2 4 6
0
50
100
150
x
y
(a) (x, y)[y = e
x
6 4 2 0 2 4 6
0
50
100
150
x
y
(b) (x, y)[y e
x
6 4 2 0 2 4 6
30
20
10
0
x
y
(c) (x, y)[y 2x x
2
; x > 0, y > 0
1 1.5 2 2.5 3 3.5 4
1
2
3
4
x
y
(d) (x, y)[xy 1; x > 0, y > 0
0 1 2 3 4 5
1
0.5
0
0.5
1
1.5
x
y
(e) (x, y)[y ln(x)
Figure 1: Sets to Exercise A1.7
3
1 Mathematical Appendix
A1.40 Sketch a few level sets for the following functions: y = x
1
x
2
, y = x
1
+ x
2
and
y = min[x
1
, x
2
].
Answer
6
-
x
1
x
2
(a) y = x
1
x
2
6
-
x
1
x
2
@
@
@
@
@
@
@
@
@
@
(b) y = x
1
+ x
2
6
-
x
1
x
2
(c) y = min(x
1
, x
2
)
Figure 2: Sets to Exercise A1.40
A1.42 Let D = [2, 2] and f : D R be y = 4 x
2
. Carefully sketch this function.
Using the denition of a concave function, prove that f is concave. Demonstrate that
the set A is a convex set.
Answer Proof of concavity: Derive the rst and second order partial derivative:
y
x
= 2x

2
y
x
2
= 2
The rst derivative is strictly positive for values x < 0 and negative for values x > 0.
The second order partial derivative is always less than zero. Therefore, the function is
concave.
Proof of convexity: The area below a concave function forms a convex set (Theorem
A1.13). Alternatively, from the denition of convexity the following inequality should
hold 4 (tx
1
+ (1 t)x
2
)
2
t(4 (x
1
)
2
) + (1 t)(4 (x
2
)
2
). Multiply out to get
4(tx
1
+x
2
tx
2
)
2
4x
2
2
+t[(x
1
)
2
(x
2
)
2
]. Again, the area below the function forms
a convex set.
6
-
x
y
Figure 3: Graph to Exercise A1.42
A1.46 Consider any linear function f(x) = a x + b for a R
n
and b R.
(a) Show that every linear function is both concave and convex, though neither is strictly
concave nor strictly convex.
Answer The statement is true i, for any x
1
, x
2
R
n
, t [0, 1], it is true that
f(tx
1
+ (1 t)x
2
) = tf(x
1
) + (1 t)f(x
2
).
4
1 Mathematical Appendix
Substituting any linear equation in this statement gives
f(tx
1
+(1t)x
2
) = a[tx
1
+(1t)x
2
]+b = tax
1
+(1t)ax
2
+tb+(1t)b = tf(x
1
)+(1t)f(x
2
)
for all x
1
, x
2
R
n
, t [0, 1].
(b) Show that every linear function is both quasiconcave and quasiconvex and, for n > 1,
neither strictly so. (There is a slight inaccuracy in the book.)
Answer As it is shown in (a) that a linear function is concave and convex, it must also
be quasiconcave and quasiconvex (Theorem A1.19). More formally, the statement
is true i, for any x
1
, x
2
R
n
(x
1
,= x
2
) and t [0, 1], we have
f(tx
1
+ (1 t)x
2
) min[f(x
1
), f(x
2
)](quasiconcavity)
f(tx
1
+ (1 t)x
2
) max[f(x
1
), f(x
2
)](quasiconvexity)
Again by substituting the equation into the denition, we get
tf(x
1
) + (1 t)f(x
2
) min[f(x
1
), f(x
2
)]
tf(x
1
) + (1 t)f(x
2
) max[f(x
1
), f(x
2
)] t [0, 1]
A1.47 Let f(x) be a concave (convex) real-valued function. Let g(t) be an increas-
ing concave (convex) function of a single variable. Show that the composite function,
h(x) = g(f(x)) is a concave (convex) function.
Answer The composition with an ane function preserves concavity (convexity). As-
sume that both functions are twice dierentiable. Then the second order partial deriva-
tive of the composite function, applying chain rule and product rule, is dened as
h

(x) = g

(f(x)) f

(x)
2
+ g

(f(x)) f

(x)
2
For any concave function,
2
f(x) 0,
2
g(x) 0, it should hold
2
h(x) 0. In
the case the two functions are convex:
2
f(x) 0 and
2
g(x) 0, it should hold

2
h(x) 0.
A1.48 Let f(x
1
, x
2
) = (x
1
5)
2
(x
2
5)
2
. Prove that f is quasiconcave.
Answer Proof: f is concave i H(x) is negative semidenite and it is strictly concave if
the Hessian is negative denite.
H =
_
2 0
0 2
_
z
T
H(x)z = 2z
2
1
2z
2
2
< 0, for z = (z
1
, z
2
) ,= 0
Alternatively, we can check the leading principal minors of H: H
1
(x) = 2 < 0 and
H
2
(x) = 4 > 0. The determinants of the Hessian alternate in sign beginning with a
negative value. Therefore, the function is even strictly concave. Since f is concave, it is
also quasiconcave.
5
1 Mathematical Appendix
A1.49 Answer each of the following questions yes or no, and justify your answer.
(a) Suppose f(x) is an increasing function of one variable. Is f(x) quasiconcave?
Answer Yes, an increasing function of one variable is quasiconcave. Any convex
combination of two points on this function will be at least as large as the smallest of
the two points. Using the dierential-based approach, f is quasiconcave, if for any
x
0
and x
1
, f(x
1
) f(x
0
) f(x
0
)/x(x
1
x
0
) 0. This must be true for any
increasing function.
(b) Suppose f(x) is a decreasing function of one variable. Is f(x) quasiconcave?
Answer Yes, a decreasing function of one variable is quasiconcave. Similarly to (a),
f is quasiconcave if for any x
0
, x
1
and t [0, 1], it is true that f(tx
0
+(1 t)x
1
)
min[f(x
0
), f(x
1
)].
(c) Suppose f(x) is a function of one variable and there is a real number b such that
f(x) is decreasing on the interval (inf, b] and increasing on [b, +inf). Is f(x)
quasiconcave?
Answer No, if f is decreasing on (inf, b] and increasing on [b, +inf) then f(x) is
not quasiconcave.
Proof: Let a < b < c, and let t
b
=
cb
ca
[0, 1], t
b
a + (1 t
b
)c = b. Given the nature
of f, f(b) < min[f(a), f(c)]. Then f(t
b
a + (1 t
b
)c) < min[f(a), f(c)], so f is not
quasiconcave.
(d) Suppose f(x) is a function of one variable and there is a real number b such that
f(x) is increasing on the interval (inf, b] and decreasing on [b, +inf). Is f(x)
quasiconcave?
Answer Yes.
Proof: Let a < b < c, for x [a, b], f(x) f(a) and for x [b, c], f(x) f(c).
Hence, for any x [a, c], f(x) min[f(a), f(c)].
(e) You should now be able to come up with a characterization of quasiconcave func-
tions of one variable involving the words increasing and decreasing.
Answer Any function of one variable f(x) is quasiconcave if and only if is either con-
tinuously increasing, continuously decreasing or rst increasing and later decreasing.
1.2 Chapter A2
A2.1 Dierentiate the following functions. State whether the function is increasing,
decreasing, or constant at the point x = 2. Classify each as locally concave, convex, or
linear at the point x = 2.
(a) f(x) = 11x
3
6x + 8 f
1
= 33x
2
6
increasing locally convex
(b) f(x) = (3x
2
x)(6x + 1) f
1
= 54x
2
6x 1
increasing locally convex
6
1 Mathematical Appendix
(c) f(x) = x
2

1
x
3
f
1
= 2x +
3
x
4
increasing locally concave
(d) f(x) = (x
2
+ 2x)
3
f
1
= (6x + 6)(x
2
+ 2x)
2
increasing locally convex
(e) f(x) = [3x/(x
3
+ 1)]
2
f
1
= 18x
x
3
3x
2
+ 1
(x
3
+ 1)
3
increasing locally concave
(f) f(x) = [(1/x
2
+ 2) (1/x 2)]
4
f
1
=
_
4
x
2

8
x
3
__
1
x
2

1
x
+ 4
_
3
increasing locally convex
(g) f(x) =
_
1
x
e
t
2
dt f
1
= e
x
2
decreasing locally convex
A2.2 Find all rst-order partial derivatives.
(a) f(x
1
, x
2
) = 2x
1
x
2
1
x
2
2
f
1
= 2 2x
1
= 2(1 x
1
) f
2
= 2x
2
(b) f(x
1
, x
2
) = x
2
1
+ 2x
2
2
4x
2
f
1
= 2x
1
f
2
= 4x
2
4
(c) f(x
1
, x
2
) = x
3
1
x
2
2
2x
2
f
1
= 3x
1
f
2
= 2(x
2
+ 1)
(d) f(x
1
, x
2
) = 4x
1
+ 2x
2
x
2
1
+ x
1
x
2
x
2
2
f
1
= 4 2x
1
+ x
2
f
2
= 2 2x
2
+ x
1
(e) f(x
1
, x
2
) = x
3
1
6x
1
x
2
+ x
3
2
f
1
= 3x
2
1
6x
2
f
2
= 3x
2
2
6x
1
(f) f(x
1
, x
2
) = 3x
2
1
x
1
x
2
+ x
2
f
1
= 6x
1
x
2
f
2
= 1 x
1
7
1 Mathematical Appendix
(g) g(x
1
, x
2
, x
3
) = ln
_
x
2
1
x
2
x
3
x
2
3
_
g
1
=
2x
1
x
2
1
x
2
x
3
x
2
3
g
2
=
x
3
x
2
1
x
2
x
3
x
2
3
g
3
=
x
2
2x
3
x
2
1
x
2
x
3
x
2
3
A2.4 Show that y = x
2
1
x
2
+ x
2
2
x
3
+ x
2
3
x
1
satises the equation
y
x
1
+
y
x
2
+
y
x
3
= (x
1
+ x
2
+ x
3
)
2
.
The rst-order partial derivatives are:
y/x
1
= 2x
1
x
2
+ x
2
3
,
y/x
2
= x
2
1
+ 2x
2
x
3
, and
y/x
3
= x
2
2
+ 2x
3
x
1
.
Summing them up gives
y
x
1
+
y
x
2
+
y
x
3
= x
2
1
+ x
2
2
+ x
2
3
+ 2x
1
x
2
+ 2x
1
x
3
+ 2x
2
x
3
= (x
1
+ x
2
+ x
3
)
2
.
A2.5 Find the Hessian matrix and construct the quadratic form, z
T
H(x)z, when
(a) y = 2x
1
x
2
1
x
2
2
H =
_
2 0
0 2
_
z
T
H(x)z = 2z
2
1
+ 2 0z
1
z
2
2z
2
2
(b) y = x
2
1
+ 2x
2
2
4x
2
H =
_
2 0
0 4
_
z
T
H(x)z = 2z
2
1
+ 2 0z
1
z
2
+ 4z
2
2
(c) y = x
3
1
x
2
2
+ 2x
2
H =
_
6x
1
0
0 2
_
z
T
H(x)z = 6x
1
z
2
1
2z
2
2
8
1 Mathematical Appendix
(d) y = 4x
1
+ 2x
2
x
2
1
+ x
1
x
2
x
2
2
H =
_
2 1
1 2
_
z
T
H(x)z = 2z
2
1
+ 2z
1
z
2
2z
2
2
(e) y = x
3
1
6x
1
x
2
x
3
2
H =
_
6x
1
6
6 6x
2
_
z
T
H(x)z = 6x
1
z
2
1
12z
1
z
2
+ 6x
2
z
2
2
A2.6 Prove that the secondorder own partial derivatives of a convex function must
always be nonnegative. Answer A convex function must be increasing over their whole
domain, i.e. might not have a maximum. Therefore, the secondorder own partial
derivative can not change from nonnegative to negative. The second-oorder own partial
derivatives represent the diagonal elements of the Hessian. Convexity of the function
requires the Hessian to be positive semidenite.
A2.7 Complete Example A2.4 for the partial with respect to x
2
. Answer The partial
with respect to x
2
is
f(x
1
, x
2
)
x
2
= Ax

1
x
1
2
= (1 )A
x
1
x
2

After multiplication by the factor t, we obtain:


f(tx
1
, tx
2
)
x
2
= (1 )A
tx
1
tx
2

= t

/t

(1 )A
x
1
x
2

=
f(x
1
, x
2
)
x
2
A2.8 Suppose f(x
1
, x
2
) =
_
x
2
1
+ x
2
2
.
(a) Show that f(x
1
, x
2
) is homogeneous of degree 1.
f(tx
1
, tx
2
) =
_
(tx
1
)
2
+ (tx
2
)
2
=
_
t
2
(x
2
1
+ x
2
2
) = t
_
x
2
1
+ x
2
2
(b) According to Eulers theorem, we should have f(x
1
, x
2
) = (f/x
1
) x
1
+(f/x
2
) x
2
.
Verify this.
1 f(x
1
, x
2
) =
x
1
_
x
2
1
+ x
2
2
x
1
+
x
2
_
x
2
1
+ x
2
2
x
2
=
x
2
1
+ x
2
2
_
x
2
1
+ x
2
2
=
_
x
2
1
+ x
2
2
9
1 Mathematical Appendix
A2.9 Suppose f(x
1
, x
2
) = (x
1
x
2
)
2
and g(x
1
, x
2
) = (x
2
1
x
2
)
3
.
(a) f(x
1
, x
2
) is homogeneous. What is its degree?
f(tx
1
, tx
2
) = t
4
(x
1
x
2
)
2
k = 4
(b) g(x
1
, x
2
) is homogeneous. What is its degree?
g(tx
1
, tx
2
) = t
9
(x
2
1
x
2
)
3
k = 9
(c) h(x
1
, x
2
) = f(x
1
, x
2
)g(x
1
, x
2
) is homogeneous. What is its degree?
h(x
1
, x
2
) = x
8
1
x
5
2
h(tx
1
, tx
2
) = t
13
(x
8
1
x
5
2
k = 13 Obviously, whenever two functions
are homogeneous of degree m and n, their product must be homogeneous of degree
m + n.
(d) k(x
1
, x
2
) = g (f(x
1
, x
2
), f(x
1
, x
2
)) is homogeneous. What is its degree?
k(tx
1
, tx
2
) = t
36
(x
1
x
2
)
18
k = 36
(e) Prove that whenever f(x
1
, x
2
) is homogeneous of degree m and g(x
1
, x
2
) is homoge-
neous of degree n, then k(x
1
, x
2
) = g (f(x
1
, x
2
), f(x
1
, x
2
)) is homogeneous of degree
mn.
k(tx
1
, tx
2
) = [t
m
(f(x
1
, x
2
), f(x
1
, x
2
))]
n
k = mn
A2.15 Check the calculations in Example A2.6 by using the substitution method to
solve the system of rstorder partials. Then evaluate the function at x

1
= 3/7 and
x

2
= 8/7 and nd y

. Verify what we found in Example A2.7 by evaluating the function


at any other point and comparing to y

. Answer The system of rstorder partials is


f(x
1
, x
2
)
x
1
= 8x
1
+ 3x
2
f(x
1
, x
2
)
x
2
= 1 + 3x
1
2x
2
Rearrange the rst partial gives: x
1
= 3/8x
2
. Substitute into the second and solve for
x
2
gives: x

2
= 8/7 and x

1
= 3/7. This is the same result as found by using the matrix
method. The value of the function at the critical point is y

(3/7, 8/7) = 4/7.


A2.16 Find the critical points when
(a) f(x
1
, x
2
) = 2x
1
x
2
1
x
2
2
. x
1
= 1 and x
2
= 0
(b) f(x
1
, x
2
) = x
2
1
+ 2x
2
1
4x
2
. x
1
= 0 and x
2
= 1
10
1 Mathematical Appendix
(c) f(x
1
, x
2
) = x
3
1
x
2
2
+ 2x
2
. x
1
= 0 and x
2
= 1
(d) f(x
1
, x
2
) = 4x
1
+ 2x
2
x
2
1
+ x
1
x
2
x
2
2
. x
1
= 10/3 and x
2
= 8/3
(e) f(x
1
, x
2
) = x
3
1
6x
1
x
2
+ x
3
2
. x
1
= 0 and x
2
= 0
A2.18 Let f(x) be a real-valued function dened on R
n
+
, and consider the matrix
H

=
_
_
_
_
_
0 f
1
f
n
f
1
f
11
f
1n
.
.
.
.
.
.
.
.
.
.
.
.
f
n
f
n1
f
nn
_
_
_
_
_
.
This is a dierent sort of bordered Hessian than we considered in the text. Here, the
matrix of second-order partials is bordered by the rstorder partials and a zero to
complete the square matrix. The principal minors of this matrix are the determinants
D
2
=

0 f
1
f
1
f
11

, D
3
=

0 f
1
f
2
f
1
f
11
f
12
f
2
f
21
f
22

, . . . , D
n
= [H

[.
Arrow & Enthoven (1961) use the sign pattern of these principal minors to establish the
following useful results:
(i) If f(x) is quasiconcave, these principal minors alternate in sign as follows: D
2
0,
D
3
0, . . . .
(ii) If for all x 0, these principal minors (which depend on x) alternate in sign
beginning with strictly negative: D
2
< 0, D
3
> 0, . . . , then f(x) is quasiconcave
on the nonnegative orthant. Further, it can be shown that if, for all x 0, we
have this same alternating sign pattern on those principal minors, then f(x) is
strictly quasiconcave on the (strictly) positive orthant.
(a) The function f(x
1
, x
2
) = x
1
x
2
+ x
1
is quasiconcave on R
2
+
. Verify that its principal
minors alternate in sign as in (ii).
Answer The bordered Hessian is
H

=
_
_
0 x
2
+ 1 x
1
x
2
+ 1 0 1
x
1
1 0
_
_
.
The two principal minors are D
2
= (x
2
+1)
2
< 0 and D
3
= 2x
1
x
2
+2x
1
0. Which
shows that the function will be quasiconcave and will be strictly quasiconcave for
all x
1
, x
2
> 0.
11
1 Mathematical Appendix
(b) Let f(x
1
, x
2
) = a ln(x
1
+x
2
) +b, where a > 0. Is this function strictly quasiconcave
for x 0? It is quasiconcave? How about for x 0? Justify.
Answer The bordered Hessian is
H

=
_
_
_
0
a
x
1
+x
2
a
x
1
+x
2
a
x
1
+x
2
a
(x
1
+x
2
)
2
a
(x
1
+x
2
)
2
a
x
1
+x
2
a
(x
1
+x
2
)
2
a
(x
1
+x
2
)
2
_
_
_
.
The two principal minors are D
2
= (
a
x
1
+x
2
)
2
< 0 for x
1
, x
2
> 0 and D
3
= 0.
Which shows that the function can not be strictly quasiconcave. However, it can be
quasiconcave following (i). For x
1
= x
2
= 0 the function is not dened. Therefore,
curvature can not be checked in this point.
A2.19 Let f(x
1
, x
2
) = (x
1
x
2
)
2
. Is f(x) concave on R
2
+
? Is it quasiconcave on R
2
+
?
Answer In order to prove concavity we can check whether the Hesssian is negative
semidenite for all x. The Hessian is
H =
_
2x
2
2
4x
1
x
2
4x
1
x
2
2x
2
1
_
.
The principal minors of the Hessian are not consistent with a concave curvature: D
1
=
2x
2
2
> 0 and D
2
= 12x
2
1
x
2
2
< 0. The bordered Hessian is
H

=
_
_
0 2x
1
x
2
2
2x
2
1
x
2
2x
1
x
2
2
2x
2
2
4x
1
x
2
2x
2
1
x
2
4x
1
x
2
2x
2
1
_
_
.
The two principal minors of the bordered Hessian are D
2
= (2x
1
x
2
)
2
< 0 and D
3
=
16x
4
1
x
4
2
> 0. Which shows that the function will be strictly quasiconcave. Note that this
function is not homogeneous of degree one.
A2.25 Solve the following problems. State the optimised value of the function at the
solution.
(a) min
x
1
,x
2
= x
2
1
+ x
2
2
s.t. x
1
x
2
= 1
x
1
= 1 and x
2
= 1 or x
1
= 1 and x
2
= 1, optimised value= 2
(b) min
x
1
,x
2
= x
1
x
2
s.t. x
2
1
+ x
2
2
= 1
x
1
=
_
1/2 and x
2
=
_
1/2 or x
1
=
_
1/2 and x
2
=
_
1/2, optimised value=
1/2
12
1 Mathematical Appendix
(c) max
x
1
,x
2
= x
1
x
2
2
s.t. x
2
1
/a
2
+ x
2
2
/b
2
= 1
x
1
=
_
a
2
/3 and x
2
=
_
2b
2
/3 or x
2
=
_
2b
2
/3, optimised value=
2ab
2
3
3
/2
(d) max
x
1
,x
2
= x
1
+ x
2
s.t. x
4
1
+ x
4
2
= 1
x
1
=
4
_
1/2 and x
2
=
4
_
1/2, optimised value=
4

2
3
= 2
3/4
(e) max
x
1
,x
2
,x
3
= x
1
x
2
2
x
3
3
s.t. x
1
+ x
2
+ x
3
= 1
x
1
= 1/6 and x
2
= 1/3 = 2/6 and x
3
= 1/2 = 3/6, optimised value= 1/432 = 108/6
6
A2.26 Graph f(x) = 6 x
2
4x. Find the point where the function achieves its
unconstrained (global) maximum and calculate the value of the function at that point.
Compare this to the value it achieves when maximized subject to the nonnegativity
constraint x 0.
Answer This function has a global optimum at x = 2. It is a maximum as the second-
order partial derivative is less than zero. The value of the function at that point is
f(2) = 10. Obviously, the global maximum is not a solution in the presence of a
nonnegativity constraint. The constrained maximization problem is
L(x, z, ) = 6 x
2
4x + (x z)
The rstorder conditions are:
L
x
= 2x 4 + = 0 (i)
L
z
= 0 (ii)
L

= x z = 0 (iii)
Since we do not know a priori whether or not the constraint will be binding, the par-
tial derivative of L with respect to the slack variable is not equal to zero. Either the
constraint is binding, i.e. the slack variable redundant z = 0, or the constraint is not
binding, i.e. the shadow value of the constraint = 0. Such a condition can be sum-
marised in stating that the product of the two variables must be zero:
zL
z
= z() = 0
If = 0, then x = 2 would solve the problem. However, it does not satisfy the
nonnegativity constraint. If ,= 0, then z = 0 and due to the equality of z and x
derived from (iii) x = 0. As the function is continuously decreasing for all values x 0,
it is the only maximizer in this range.
13
1 Mathematical Appendix
6
-
x
y
Figure 4: Graph to Exercise A2.26
14
2 Consumer Theory
2 Consumer Theory
2.1 Preferences and Utility
1.6 Cite a credible example were the preferences of an ordinary consumer would be
unlikely to satisfy the axiom of convexity.
Answer: Indierence curves representing satiated preferences dont satisfy the axiom of
convexity. That is, reducing consumption would result in a higher utility level. Negative
utility from consumption of bads (too much alcohol, drugs, unhealthy food etc.) would
rather result in concave preferences.
1.8 Sketch a map of indierence sets that are parallel, negatively sloped straight lines,
with preference increasing northeasterly. We know that preferences such as these satisfy
Axioms 1, 2, 3, and 4. Prove the they also satisfy Axiom 5

. Prove that they do not


satisfy Axiom 5.
Answer: Denition of convexity (Axiom 5

): If x
1
x
0
, then tx
1
+ (1 t)x
0
x
0
for
all t [0, 1]. Strict convexity (Axiom 5) requires that, if x
1
,= x
0
and x
1
x
0
, then
tx
1
+ (1 t)x
0
~ x
0
for all t [0, 1].
The map of indierence sets in Figure 5 represents perfect substitues. We know that
those preferences are convex but not stricly convex. Intuitively, all combinations of two
randomly chosen bundles from one indierence curve will necessarily lie on the same
indierence curve. Additionally, the marginal rate of substitution does not change by
moving from x
0
to x
1
. To prove the statement more formally, dene x
t
as convex
combination of bundles x
0
to x
1
: x
t
= tx
0
+ (1 t)x
1
. Re-writing in terms of single
commodities gives us:
x
t
= (tx
0
1
, tx
0
2
) +
_
(1 t)x
1
1
, (1 t)x
1
2
_
.
A little rearrangement and equalising the two denitions results in the equality
tx
0
+ (1 t)x
1
= (tx
0
1
+ (1 t)x
1
1
), (tx
0
2
+ (1 t)x
1
2
).
That is, the consumer is indierent with respect to the convex combination and the
original bundles, a clear violation of strict convexity.
15
2 Consumer Theory
1 2 3 4 5 6 7
2
4
6
x
1
x
2
Figure 5: Indierence sets to Exercise 1.8
1.9 Sketch a map of indierence sets that are parallel right angles that kink on the
line x
1
= x
2
. If preference increases northeasterly, these preferences will satisfy Axioms
1, 2, 3, and 4. Prove that they also satisfy Axiom 5. Do they also satisfy Axiom 4?
Do they satisfy Axiom 5?
Answer: Convexity (Axiom 5

) requires that, if x
1
x
0
, then tx
1
+ (1 t)x
0
x
0
for
all t [0, 1].
Take any two vectors x
0
, x
1
such that x
0
x
1
. Given the nature of these preferences,
it must be true that min[x
0
1
, x
0
2
] = min[x
1
1
, x
1
2
]. For any t [0, 1] consider the point
tx
1
+ (1 t)x
2
. If we can show that
min[tx
0
1
+ (1 t)x
0
2
, tx
1
1
+ (1 t)x
1
2
] min[x
0
1
, x
0
2
] = min[x
1
1
, x
1
2
],
then we have shown that these preferences are convex.
min[tx
0
1
+ (1 t)x
0
2
, tx
1
1
+ (1 t)x
1
2
] min[tx
0
1
, tx
1
1
] + min[(1 t)x
0
2
, +(1 t)x
1
2
]
= min[x
0
2
, x
1
2
] + t[min(x
0
1
, x
1
1
) min(x
0
2
, x
1
2
)] = min[x
0
2
, x
1
2
]
Denition of strict monotonicity (Axiom 4): For all x
0
, x
1
R
n
+
, if x
0
x
1
, then
x
0
x
1
, while if x
0
x
1
, then x
0
~ x
1
.
The map of indierence sets in Figure 6 represents perfect complements. Take two points
x
0
, x
1
along one indierence curve. If x
0
x
1
, preferences increase northeasterly,
then x
0
~ x
1
. For any two vectors on the same indierence curve, that is x
0
x
1
, it
follows x
0
x
1
. Therefore, the denition of strict monotonicity is satised for these
indierence sets.
Strict convexity (Axiom 5) requires that, if x
1
,= x
0
and x
1
x
0
, then tx
1
+(1t)x
0
~ x
0
for all t [0, 1].
Take any two points along the horizontal or vertical part of an indierence curve such
as (x
0
1
, x
0
2
) and (x
0
1
, x
1
2
), where x
0
2
> x
1
2
. Any convex combination x
t
= x
0
1
, tx
0
2
+(1 t)x
1
2
lies on the same indierence curve as x
1
and x
0
. Therefore, it is not possible that
16
2 Consumer Theory
x
t
~ tx
0
+ (1 t)x
1
. That is, the consumer is indierent with respect to the convex
combination and the original bundles, a clear violation of strict convexity.
6
-
x
1
x
2
p
x
0
px
1
Figure 6: Indierence sets to Exercise 1.9
1.12 Suppose u(x
1
, x
2
) and v(x
1
, x
2
) are utility functions.
(a) Prove that if u(x
1
, x
2
) and v(x
1
, x
2
) are both homogeneous of degree r, then s(x
1
, x
2
)
u(x
1
, x
2
) + v(x
1
, x
2
) is homogeneous of degree r.
Answer: Whenever it holds that t
r
u(x
1
, x
2
) = u(tx
1
, tx
2
) and t
r
v(x
1
, x
2
) = v(tx
1
, tx
2
)
for all r > 0, it must also hold that t
r
s(x
1
, x
2
) u(tx
1
, tx
2
) + v(tx
1
, tx
2
) =
t
r
u(x
1
, x
2
) + t
r
v(x
1
, x
2
).
(b) Prove that if u(x
1
, x
2
) and v(x
1
, x
2
) are quasiconcave, then m(x
1
, x
2
) minu(x
1
, x
2
), v(x
1
, x
2
)
is also quasiconcave.
Answer: Forming a convex combination of the two functions u and v and comparing
with m(x
t
) satises the denition of quasiconcavity:
When u(x
t
) min
_
tu(x
1
) + (1 t)u(x
2
)
_
and
v(x
t
) min
_
tv(x
1
) + (1 t)v(x
2
)
_
so
m(x
t
) min
_
u(x
t
), v(x
t
)
_
.
2.2 The Consumers Problem
1.20 Suppose preferences are represented by the Cobb-Douglas utility function, u(x
1
, x
2
) =
Ax

1
x
1
2
, 0 < < 1, and A > 0. Assuming an interior solution, solve for the Marshal-
lian demand functions.
Answer: Use either the Lagrangian or the equality of Marginal Rate of Substitution and
price ratio. The Lagrangian is
L = Ax

1
x
1
2
+ (y p
1
x
1
p
2
x
2
). The rst-order conditions (FOC) are
L
x
1
= Ax
1
1
x
1
2
p
1
= 0
L
x
2
= (1 )Ax

1
x

2
p
2
= 0
L

= y p
1
x
1
+ p
2
x
2
= 0
17
2 Consumer Theory
By dividing rst and second FOC and some rearrangement, we get either x
1
=
x
2
p
2
(1)p
1
or
x
2
=
(1)p
1
x
1
p
2
. Substituting one of these expressions into the budget constraint, results
in the Marshallian demand functions: x
1
=
y
p
1
and x
2
=
(1)y
p
2
.
1.21 Weve noted that u(x) is invariant to positive monotonic transforms. One com-
mon transformation is the logarithmic transform, ln(u(x)). Take the logarithmic trans-
form of the utility function in 1.20; then, using that as the utility function, derive the
Marshallian demand functions and verify that they are identical to those derived in the
preceding exercise (1.20).
Answer: Either the Lagrangian is used or the equality of Marginal Rate of Substitution
with the price ratio. The Lagrangian is
L = ln(A) + ln(x
1
) + (1 ) ln(x
2
) + (y p
1
x
1
p
2
x
2
). The FOC are
L
x
1
=

x
1
p
1
= 0
L
x
2
=
(1 )
x
2
p
2
= 0
L

= y p
1
x
1
+ p
2
x
2
= 0
The Marshallian demand functions are: x
1
=
y
p
1
and x
2
=
(1)y
p
2
. They are exactly
identical to the demand functions derived in the preceding exercise.
1.22 We can generalise further the result of the preceding exercise. Suppose that
preferences are represented by the utility function u(x). Assuming an interior solution,
the consumers demand functions, x(p, y), are determined implicitly by the conditions
in (1.10). Now consider the utility function f(u(x)), where f/ > 0, and show that the
rstorder conditions characterising the solution to the consumers problem in both cases
can be reduced to the same set of equations. Conclude from this that the consumers
demand behavior is invariant to positive monotonic transforms of the utility function.
Answer: The set of rstorder conditions for a utility function and its positive monotone
transform, assuming a linear budget constraint, are as follows:
L
x
i
=
u
x
i
p
i
L
x
i
=
f(u)
u
u(x)
x
i
p
i
L
x
j
=
u
x
j
p
j
L
x
j
=
f(u)
u
u(x)
x
j
p
j
u/x
i
u/x
j
=
p
i
p
j
f(u)/u
f(u)/u
u/x
i
u/x
j
=
p
i
p
j
After forming the Marginal Rate of Substitution, the outer derivative cancels out. There-
fore, the demand function should be unaected by the positive monotonic transformation
of the utility function.
18
2 Consumer Theory
1.24 Let u(x) represent some consumers monotonic preferences over
x R
n
+
. For each of the functions F(x) that follow, state whether or not f also represents
the preferences of this consumer. In each case, be sure to justify your answer with either
an argument or a counterexample.
Answer:
(a) f(x) = u(x) + (u(x))
3
Yes, all arguments of the function u are transformed equally
by the third power. Checking the rst- and second-order partial derivatives reveals
that, although the second-order partial is not zero, the sign of the derivatives is
always invariant and positive.

2
f
x
2
i
=

2
u
x
2
i
+ 6(u(x))
u
x
i
+ 3(u(x)
2

2
u
x
2
i
Thus, f represents a monotonic transformation of u.
(b) f(x) = u(x) (u(x))
2
No, function f is decreasing with increasing consumption for
any u(x) < (u(x))
2
. Therefore, it can not represent the preferences of the consumer.
It could do so if the minus sign is replaced by a plus sign.
(c) f(x) = u(x) +

n
i=1
x
i
Yes, the transformation is a linear one, as the rst partial
is a positive constant, here one, and the second partial of the transforming function
is zero. Checking the partial derivatives proves this statement:
f
x
i
=
u
x
i
+ 1 and

2
f
x
2
i
=

2
u
x
2
i
.
1.28 An innitely lived agent owns 1 unit of a commodity that she consumes over her
lifetime. The commodity is perfect storable and she will receive no more than she has
now. Consumption of the commodity in period t is denoted x
t
, and her lifetime utility
function is given by
u(x
0
, x
1
, x
2
, . . .) =

t=0

t
ln(x
t
), where 0 < < 1.
Calculate her optimal level of consumption in each period.
Answer: Establish a geometric series to calculate her lifetime utility:
u =
0
ln(x
0
) + ln(x
1
) +
2
ln(x
2
) + . . . +
t
ln(x
t
)
As is less than one, this series approaches a nite value. To nd the solution, multiply
the expression by and subtract from the original equation [(1)-(2)].
u =
1
ln(x
0
) +
2
ln(x
1
) +
3
ln(x
2
) + . . . +
t+1
ln(x
t
)
u u = (1 )u = ln(x
0
)
t+1
ln(x
t
)
u =
ln(x
0
)
t+1
ln(x
t
)
1
= ln(x
0
)
Thus, the consumers utility maximising consumption will be constant in every period.
19
2 Consumer Theory
2.3 Indirect Utility and Expenditure
1.29 In the two-ggod case, the level sets of the indirect utility function in price space
are sets of the form (p
1
, p
2
)[v(p
1
, p
2
, y) = v
0
for v
0
R. These are sometimes called
priceindierence curves. Sketch a possible map of priceindierence curves. Give sep-
arate arguments to support your claims as to their slope, curvature, and the direction
of increasing utility.
Answer: Figure 7 presents a map of priceindierence curves. Two budget constraints
represent dierent price vectors holding income and utility constant. The combination
of prices can be recovered by moving along the priceindierence curve. Because v is
decreasing in prices, utility increases by moving further towards the origin. The price
indierence curve must be stricly convex towards the origin. The priceindierence
curves must be negatively sloped, i.e. one of the prices must decrease whenever the other
price increases holding utility constant. In order to illustrate this fact, re-arrange a ho-
mothetic indirect utility function in a twocommodity world v(p, y) = (y)v(f(p
1
, p
2
, 1)
in terms of p
1
and dierentiate with respect to p
2
.
p
1
= v(y)
1
f(p
1
)
1
p
1
p
2
= v(y)
1
f(p
1
)
2
f(p
1
)
p
1
Figure 7: Map of priceindierence curves
1.30 Show that the indirect utility function in Example 1.2 is a quasi-convex function
of prices and income.
20
2 Consumer Theory
Answer: The indirect utility function corresponding to CES preferences is: v(p, y) =
y (p
r
1
+ p
r
2
)
1/r
, where r /( 1).
There are several ways. First, using the inequality relationship, let p
t
= tp
0
+(1 t)p
1
and y
t
= ty
0
+ (1 t)y
1
. We need to show that the indirect utility function fullls the
inequality
y
_
p
tr
1
+ p
tr
2
_
1/r
max[y
_
p
0r
1
+ p
0r
2
_
1/r
, y
_
p
1r
1
+ p
1r
2
_
1/r
]
which gives:
y
_
t
r
(p
0r
1
+ p
0r
2
) + (1 t)
r
(p
1r
1
+ p
1r
2
)
_
1/r
max[y
_
p
0r
1
+ p
0r
2
_
1/r
, y
_
p
1r
1
+ p
1r
2
_
1/r
].
Remember that the inverse of concave function is convex. Therefore, it is sucient to
prove that, for any v R and y > 0, the set p R
2
: v(p, y) v is convex. Dene
f(p) = (p
r
1
+ p
r
2
)
1/r
. If r (0, 1), then f(p)
r
= p
r
1
+ p
r
2
is a concave function. Hence
f(p) = (f(p)
1/r
) is convex. Since v(p, y) = y/f(p), this implies that the indirect utility
funtion is convex for every y and p.
1.37 Verify that the expenditure function obtained from the CES direct utility function
in Example 1.3 satises all the properties given in Theorem 1.7.
Answer: The expenditure function in a twocommodity world is e(p, u) = u(p
r
1
+ p
r
2
)
1/r
where r /( 1).
1. Zero when u takes on the lowest level of utility in U.
The lowest value in U is u((0)) because the utility function is strictly increasing.
Consequently, 0(p
r
1
+ p
r
2
)
1/r
= 0.
2. Continuous on its domain R
n
++
U.
This property follows from the Theorem of Maximum. As the CES direct utility
function satises the axiom of continuity, the derived expenditure function will be
continuous too.
3. For all p >> 0, strictly increasing and unbounded above in u.
Take the rst partial derivative of the expenditure function with respect to utility:
e/u = (p
r
1
+p
r
2
)
1/r
. For all strictly positive prices, this expression will be positive.
Alternatively, by the Envelope theorem it is shown that the partial derivative of the
minimum-value function e with respect to u is equal to the partial derivative of the
Lagrangian with respect to u, evaluated at (x

), what equals . Unboundness


above follows from the functional form of u.
4. Increasing in p.
Again, take all rst partial derivatives with respect to prices: e/p
i
= up
r1
i
(p
r
1
+ p
r
2
)
1
r
1
,
what is, obviously, positive.
5. Homogeneous of degree 1 in p.
e(tp, u) = u((tp
1
)
r
+ (tp
2
)
r
)
1/r
= t
1
u(p
r
1
+ p
r
2
)
1/r
21
2 Consumer Theory
6. Concave in p.
The denition of concavity in prices requires
t
_
u
_
p
0
r
1
+ p
0
r
2
_
1/r
_
+ (1 t)
_
u
_
p
1
r
1
+ p
1
r
2
_
1/r
_
e(p
t
, u)
for p
t
= tp
0
+(1 t)p
1
. Plugging in the denition of the price vector into e(p
t
, u)
yields the relationship
t
_
u
_
p
0
r
1
+ p
0
r
2
_
1/r
_
+ (1 t)
_
u
_
p
1
r
1
+ p
1
r
2
_
1/r
_

u
_
t(p
0
r
1
+ p
0
r
2
) + (1 t)(p
1
r
1
+ p
1
r
2
)
_
1/r
.
Alternatively, we can check the negative semideniteness of the associated Hessian
matrix of all second-order partial derivatives of the expenditure function. A third
possibility is to check (product rule!)

2
e
p
2
i
= u
_
(r 1)p
r2
i
(p
r
1
+ p
r
2
)
1/r1
rp
r1
i
(p
r
1
+ p
r
2
)
1/r2
_
< 0 by r < 0.
7. Shephards lemma
e/p
1
= up
r1
1
(p
r
1
+p
r
2
)
1/r1
what is exactly the denition of a CES-type Hicksian
demand function.
1.38 Complete the proof of Theorem 1.9 by showing that
x
h
(p, u) = x (p, e(p, u)).
Answer: We know that at the solution of the utility maximisation or expenditure min-
imisation problem e(p, u) = y and u = v(p, y). Substitute the indirect utility function
v into the Hicksian demand function gives x
h
(p, v(p, y)). As the new function is a
function of prices and income only, it is identical to the Marshallian demand function.
Furthermore, by replacing income by the expenditure function we get the expression
x (p, e(p, u)).
2.4 Properties of Consumer Demand
1.40 Prove that Hicksian demands are homogeneous of degree zero in prices.
Answer: We know that the expenditure function must be homogeneous of degree one
in prices. Because any Hicksian demand function equals, due to Shephards lemma, the
rst partial derivative of the expenditure function and, additionally, we know that the
derivatives degree of homogeneity is k 1. The Hicksian demand functions must be
homogeneous of degree 1 1 = 0 in prices.
1.42 For expositional purposes, we derived Theorems 1.14 and 1.15 separately, but
really the second one implies the rst. Show that when the substitution matrix (p, u)
is negative semidenite, all ownsubstitution terms will be nonpositive.
22
2 Consumer Theory
Answer: Theorem 1.15 implies Theorem 1.14 because all elements of the substitution
matrix represent secondorder partial derivatives of the expenditure function. Therefore,
the secondorder cross-partial derivatives must be equal due to Youngs theorem which
results in a symmetric matrix of secondorder partial derivatives. The diagonal elements
of this matrix must be negative because the expenditure function is required to be
concave in prices. Subsequently, any compensated demand function is required to be
nonincreasing in its own price. Thus, the leading principal minors will be alternating
in sign, starting with negative:
(1)
1
D
1
=

2
e(p, u)
p
2
1
=
x
h
1
(p, u)
p
1
0
(1)
2
D
2
=

2
e(p, u)
p
2
1

2
e(p, u)
p
2
2

2
e(p, u)
p
1
p
2
_
2
0
(1)
3
D
3
0
1.43 In a two-good case, show that if one good is inferior, the other good must be
normal.
Answer: The Engel-aggregation in a two-good case is the product of the income elasticity
and the repsective expenditure share s
1

1
+ s
2

2
= 1. An inferior good is characterised
by a negative income elasticity, thus, one of the two summands will be less than zero.
Therefore, to secure this aggregation, the other summand must be positive (even larger
one) and the other commodity must be a normal good (even a luxury item).
1.55 What restrictions must the
i
, f(y), w(p
1
, p
2
), and z(p
1
, p
2
) satisfy if each of the
following is to be a legitimate indirect utility function?
Answer:
(a) v(p
1
, p
2
, p
3
, y) = f(y)p

1
1
p

2
2
p

3
3
The function f(y) must be continuous, strictly in-
creasing and homogeneous of degree 0

i
. Each of the exponents
i
has to
be less than zero to satisfy v decreasing in prices. Furthermore, negative partial
derivatives of v with respect to each price are required to get positive Marshallian
demand functions by using Roys identity.
(b) v(p
1
, p
2
, y) = w(p
1
, p
2
)+z(p
1
, p
2
)/y Both functions, w and z, must be continuous and
decreasing in prices. Function z has to be homogeneous of degree one and function
w has to be homogeneous of degree zero:
v(tp
1
, tp
2
, ty) = t
0
w(p
1
, p
2
) + (t
1
z(p
1
, p
2
))/(ty) = t
0
(w(p
1
, p
2
) + z(p
1
, p
2
)/y) .
To satisfy v increasing in income, function z must be < 0.
1.60 Show that the Slutsky relation can be expressed in elasticity form as
ij
=
h
ij

s
j

i
, where
h
ij
is the elasticity of the Hicksian demand for x
i
with respect to price p
j
,
23
2 Consumer Theory
and all other terms are as dened in Denition 1.6.
Answer: The Slutsky relation is given by
x
i
p
j
=
x
h
i
p
j
x
j
x
i
y
.
Multiplying the total expression with y/y and p
j
gives
x
i
p
j
p
j
=
x
h
i
p
j
p
j

p
j
x
j
y
x
i
y
y.
By assuming that x
h
i
= x
i
before the price change occurs, we can divide all three terms
by x
i
. The result of this operation is
x
i
p
j
p
j
x
i
=
x
h
i
p
j
p
j
x
i
s
j
x
i
y
y
x
i
=
ij
=
h
ij
s
j

i
1.62 The substitution matrix of a utilitymaximising consumers demand system at
prices (8, p) is
_
a b
2 1/2
_
. Find a, b, and p.
Answer: Whenever demand is generated from utility maximisation, the substitution
matrix must possess the followign three properties: negative semideniteness, symmetry,
and satisfy (p, u)p = 0 (known as Hicks Third Law). Putting things together, due
to symmetry b = 2. From the second row we get p
2
= 32 because 2p
1
1/2p
2
= 0.
Subsequently, a = 8. Checking the rst and second leading principal minor to see
whether the matrix is negative semidenite, gives D
1
= 8 < 0 and D
2
= 4 4 = 0.
Thus, the matrix is negative semidenite.
Additional exercise Relationship between utility maximisation and expenditure min-
imisation
Lets explore the relationship with an example of a concrete utility function. A con-
sumers utility function is u = x
1/2
1
x
1/2
2
. For the derived functions see 1
2.5 Equilibrium and Welfare
4.19 A consumer has preferences over the single good x and all other goods m repre-
sented by the utility function, u(x, m) = ln(x) +m. Let the price of x be p, the price of
m be unity, and let income be y.
(a) Derive the Marshallian demands for x and m.
Answer The equality of marginal rate of substitution and price ratio gives 1/x = p.
Thus, the Marshallian demand for x is x = 1/p. The uncompensated demand for m
separates into two cases depending on the amount of income available:
m =
_
0 when y 1
y 1 when y > 1.
24
2 Consumer Theory
Start from the utility function Minimise expenditures s.t. u
and derive the Marshallian demand for x
1
to nd the Hicksian demand function
x
1
= y/2p
1
x
h
1
= u(p
2
/p
1
)
1/2
Plug in the respective demand functions to get the
indirect utility function expenditure function
v = y/(4p
1
p
2
)
1/2
e = u(4p
1
p
2
)
1/2
Substitute the expenditure function Substitute the indirect utility function
into the Marshallian demand function into the Hicksian demand function
to derive the Hicksian demand function to derive the Marshallian demand function
x
1
= (u(4p
1
p
2
)
1/2
)/2p
1
= u(p
2
/p
1
)
1/2
x
h
1
= (p
2
/p
1
)
1/2
y/(4p
1
p
2
)
1/2
= y/2p
1
Invert v, rearrange and replace v by u Invert e, rearrange and replace e by y
to get the expenditure function to get the indirect utility function
e = y = u(4p
1
p
2
)
1/2
v = u = y(4p
1
p
2
)
1/2
Check Roys identity Check Shephards lemma

v/p
1
v/y
=
2y(p
1
p
2
)
1/2
4(p
3
1
p
2
)
1/2
= y/2p
1
e
p
1
=
u4p
2
2(4p
1
p
2
)
1/2
= u(p
2
/p
1
)
1/2
Establish the Slutsky equation
x
1
p
2
=
u
2(p
1
p
2
)
1/2

y
2p
2

1
2p
1
substitute u = v(p, y) into the substitution eect
x
1
p
2
=
y
4p
1
p
2

y
4p
1
p
2
= 0
Table 1: Relationship between UMP and EMP
(b) Derive the indirect utility function, v(p, y).
Answer Again, depending on the amount of income available there will be two
indirect utility functions:
v(p, y) =
_
ln
_
1
p
_
when m 1
y 1 ln p when m > 1.
(c) Use the Slutsky equation to decompose the eect of an own-price change on the
demand for x into an income and substitution eect. Interpret your result briey.
Answer A well-known property of any demand function derived from a quasi-linear
utility function is the absence of the income eect. Which can be easily seen in the
application of the Slutsky equation:
x
p
=
x
h
p
+ x
x
y
x
p
=
1
p
2
+ 0
1
p
=
x
h
p
Therefore, the eect of an own-price change on the demand for x consists of the
substitution eect only, the partial derivative of the compensated demand function
with respect to price.
(d) Suppose that the price of x rises from p
0
to p
1
> p
0
. Show that the consumer surplus
area between p
0
and p
1
gives an exact measure of the eect of the price change on
25
2 Consumer Theory
Figure 8: Graph to 4.19
consumer welfare.
Answer The consumer surplus area can be calculated by integrating over the inverse
uncompensated demand function of x:
CS =
_
p
0
p
1
1
x
dx = ln p
0
ln p
1
.
Calculating the change in utility induced by a price change gives:
v = v
1
(p
1
, y
0
) v
0
(p
0
, y
0
) = y 1 ln p
1
(y 1 ln p
0
) = ln p
1
+ ln p
0
.
As the two expressions are equal, the consumer surplus area gives an exact measure
of the eect of the price change on consumer welfare in the case of quasi-linear
preferences.
(e) Carefully illustrate your ndings with a set of two diagrams: one giving the indif-
ference curves and budget constraints on top, and the other giving the Marshallian
and Hicksian demands below. Be certain that your diagrams reect all qualitative
information on preferences and demands that youve uncovered. Be sure to consider
the two prices p
0
and p
1
, and identify the Hicksian and Marshallian demands.
Answer See Figure 8. Please note, that Hicksian and Marshallian demands are
identical here.
26
References
References
Arrow, K. J. & Enthoven, A. C. (1961), Quasi-concave programming, Econometrica
29(4), 779800.
27