Solutions Manual for

Advanced Microeconomic Theory I

Jianfei Shen

School of Economics, Shanghai University of Finance and Economics

E-mail address : jianfeishen@163.com

URL: http://www.shufe.edu.cn
http://www.shufe.edu.cn
Thanks.
Thanks.
 Contents

Part
Part 1. Consum
Consumer
er Theory
Theory 1
Chapter 1. Solution for Exercise I 3
1. (JR, 1.4) 3
2. (JR, 1.27) 3
3. (JR, 1.56) 4
4. (JR, 1.24) 5
5. (Homogeneity) 6
Chapter 2. Solution for Exercise II 9
1. (JR, 1.36) 9
2. (JR, 2.10) 9
3. (MWG, 2.F.7) 10
4. (JR, 1.62) 11
5. (JR, 2.9) 11

Part
Part 2. Production
Production and Choice under Uncert
Uncertaint
ainty
y 13
Chapter 3. Solution for Exercise III 15
1. (JR, 3.36) 15
2. (JR, 3.39) 16
3. (The Allais Paradox) 17
4. (The Simplex) 18
5. (Independence Axiom) 19
Chapter 4. Solution for Exercise IV 21
1. (MWG, 6.C.16) 21
2. (MWG, 6.C.20) 22
3. (Betweenness Axiom) 23
4. (Quadratic v.N-M Utility Function) 24

i
 Part 1

Consumer Theory
 CHAPTER 1

Solution for Exercise I

1. (JR, 1.4)
L
Proof.  Suppose that  x,y,z ∈ R + , and x  y, y  z. By   Definition 1.21,
we have
(1.1) x   y  ⇔  x  y ∧ y   x;
and
(1.2) y    z  ⇔  y   z ∧ z   y.
The transitivity implies that  x  z. Suppose that z   x. Since y   z [by equation
(1.2)], the transitivity then implies that y  x. But this contradicts (1.1). Thus
z   x. Hence
x   z.
Suppose x  ∼  y, and y  ∼  z. By  Definition 1.3 2, we have
(1.3) x ∼  y  ⇔  x  y ∧ y   x;
and
(1.4) y  ∼  z  ⇔  y   z ∧ z   y.
The transitivity implies that
xz and z   x.
Hence
x ∼  z.


2. (JR, 1.27)
Solution. If  x 1  < x2 ,
u(x) = max[ax1 , ax2 ] + min[x1 , x2 ] = ax 2  + x1 ;
If  x 2  < x1 ,
u(x) = max[ax1 , ax2 ] + min[x1 , x2 ] = ax 1  + x2 .

p1
Case i: a <  p2
< a1 . In this case,
w
x1  = x 2  = .
 p1  + p2
1Jehle & Reny, 2001, P6
2Jehle & Reny, 2001, P7

3
4 1. SOLUTION FOR EXERCISE I

x2

C 

O B x1

Figure 2.1.   The Indifference Curve

p1
Case ii:  p2
> a1 . In this case, there is a corner solution
w
x1  = 0, x2  = .
 p2
p1
Case iii:  p2
< a. In this case, there is a corner solution, too
w
x1  = , x2  = 0.
 p1
p1
Case iv:  p2
= a1 . In this case, the UMP is

x1  =
w − p2 x2
, x2  ∈
w
,
w
.

 p1  p1  + p2  p2
p1
Case v:  p2
= a. In this case, the UMP is

x1  =
w − p2 x2
, x2  ∈ 0,
w
.

 p1  p1  + p2


3. (JR, 1.56)
Solution.  Use the monotone transformation of the given utility function 3:
 n
v(x) = ln u(x) = bi ln(xi  − ai ).
i=1

The UMP can written as

n
max bi ln(xi − ai );
i=1
 n
s.t.  pi xi  ≤  w.
i=1

3Stone, J.E. (1954): Linear Expenditure Systems and Demand Analysis: An Application to
the Pattern of British Demand.  Economic Journal   64: 511–27
 4. (JR, 1.24) 5

Then form the Lagrangian, we obtain

n
    n
L   = bi ln(xi − ai ) + λ w −  pi xi .
i=1 i=1

The first-order condition of the UMP yields

∂ L   bi
(1.5) = − λpi  = 0, ∀i.
∂x i xi − ai
Summarize all of the first-order conditions:
n
 1
(1.6) 1 = λ  pi (xi − ai )  ⇒  λ =  n .
i=1 i=1 pi (xi  − ai )
With (1.5) and (1.6), we have
pi (xi − ai ) pi (xi − ai )
bi  =  n =  n .
i=1 pi (xi  − ai ) w − i=1 pi ai
So  n
 b i (w − i=1 pi ai )
xi  = a i + , ∀i.
 pi


4. (JR, 1.24)
Solution.   A utility function that represents a preference relation    is not
unique. For any  strictly increasing function : f  :  R  →  R , v(x) = f (u(x)) is a new
utility function representing the same preferences as  u(·)4.
a: f (x) =  u(x) + [u(x)]2 .
∂f (·)
 = 1 + 3[u(x)]2 > 0.
∂u(·)
So this is a strictly increasing function, and it represents the same  .
b: f (x) =  u(x) + [u(x)]2 .
∂f (·)
 = 1 + 2u(x).
∂u(·)
∂f (·)
• ≤ 0 if  u(x) ≤ − 12 . In this case, f (x)   does not   represents the
∂u (·)
same   ;
(·)
• ∂f 
∂u (·)
> 0 if  u(x) >  − 12 . In this case,  f (x) represents the same   .
c: f (x) =  u(x) + ni=1 xi .

Not necessarily. f  represents the same preferences as u if  u(x) is a
n 
monotonic transformation  of  i=1 xi . Otherwise, they may not represent
the same preferences. For example, suppose  u(x) =  x1 . Then
u(1, 0, . . . , 0) > u(0, 3, 0, . . . , 0),
but
f (1, 0, . . . , 0) = 1 + 1 = 2  < 3 = 0 + 3 =  f (0, 3, 0, . . . , 0).

4Let x, y ∈ R L . Since u(·) represents , x  y iff  u (x) ≥ u(y ). Since f (·) is strictly
+
increasing,  u (x)  ≥  u (y ) iff  v (x)  ≥  v (y). Hence  x  y iff  v (x)  ≥  v (y ). Therefore  v (·) represents  .
This completes the proof.
6 1. SOLUTION FOR EXERCISE I

5. (Homogeneity)
L L
Proof. a:   We want to show that  ∀ p  ∈  R ++ , y  ≥  0, α  ≥  0, and x  ∈  R + , if  x
is the Marshallian demand at ( p, y), then x = αx is the Marshallian demand at
( p, αy). We’ll complete this proof with two steps:
Step I: we prove that the consumption bundle  x is affordable at  ( p, αy).

Note that p · x  ≤  y  ⇒  p · αy  ≤  αy, so x  = αx is affordable at ( p, αy) indeed.

Step II: we prove that the consumption bundle  x is the Marshallian demand 
 function at  ( p, αy).
L
Let x  ∈  R + and p · x ≤  αy. Then
 x 
 p · ≤  y.
α
Hence
x
u
   < u(x), (by the Strictly quasiconcave)
α
since x  is the Marshallian demand function at ( p, y).
Thus, by the homogeneity of  u(·):
x
u
  1
 = u (x ) < u(x) ⇒  u(x ) ≤  αu(x) =  u(αx).
α α
Hence αx  is the Marshiallian demand function at ( p, αy), and
x( p, αy) =  αx( p, y).
b:  By the result of part (a), we have
v( p, αy) =  u(x( p, αy)) = u(α · x( p, y)) = α · u(x( p, y)) = α · v( p, y).
This proves that the indirect utility function  v (·) is homogeneous of degree one in
y.
L
c:  In this part, we’ll show that  ∀ x  ∈  R + and α  ≥  0, we have  u(αx) =  αu(x).
We first prove that
x( p, αy) =  α · x( p, y),
where x(·) is the Marshiallian demand function.
Since v (·, ·) is homogeneous of degree one in  y:
v( p, αy) =  αv( p, y),
we have
∇ p v( p, αy) =  α∇ p v( p, y),
and
∇y v( p, αy) =  α · ∇y v( p, αy) =  α · ∇y v( p, y)  ⇒ ∇y v( p, αy) =  ∇ y v( p, y).
Then by Roy’s identity:
∂v( p, y)/∂p i
xi ( p, y) =  −  , i = 1, . . . , L ,
∂v( p, y)/∂y
we have
∇ p v( p, αy)  ∇ p v( p, y)
x( p, αy) =  − =  −α · = α · x( p, y).
∇y v( p, αy) ∇y v( p, y)
 5. (HOMOGENEITY) 7

Now, again by the homogeneity of  v (·, ·), we have

u(α · x( p, y)) =  u(x( p, αy)) = v( p, αy) =  α · v( p, y) =  α · u(x( p, y)).
This completes the proof of 
u(α · x( p, y)) =  α · u(x( p, y)).

 CHAPTER 2

Solution for Exercise II

1. (JR, 1.36)
Solution. a.   By definition,

e( p, u)  min  p · x
L
x∈R+

s.t. u(x) ≥  u
The solution to the EMP is known as the  Hicksian demand function : h( p, u). Hence
u(x0 )  ≥  u 0 ,
and so
e( p, u0 )  ≤  p · x0 ,
with the equality sign holds when  p = p 0 .
b.  It follows immediately from part a that
f ( p) =  e( p, u) − p · x0
is maximized at  p = p0 :
max f ( p) = 0.
c. If  f ( p) is differentiable at  p 0 ,
∇ p f   p0  = 0.
(2.1)

d. If  e( p, u) is differentiable in  p, then by equation (2.1) we have
∇ p f  =  ∇ p e − x0 = 0,
so
h( p, u) =  ∇ p e( p, u).


2. (JR, 2.10)
Solution. a.

Bundle
0
x x1 x2
 p0 42   48 40∗
Price p1 33∗ 36 39(∗)
 p2 52 48∗ 51
9
10 2. SOLUTION FOR EXERCISE II

These consumption bundles are plausible in that the WARP is satisfied. In partic-
ular, we note
 p1 x1 = 36  > 33 = p1 x0 ,
 p0 x0 = 42  < 48 = p0 x1 ,
so
(2.2) x1 R x0 .
 p0 x0 = 42  > 40 = p0 x2 ,
 p2 x2 = 50  < 52 = p2 x0 ,
so
(2.3) x0 R x2 .
 p2 x2 = 50  > 48 = p2 x1 ,
 p1 x1 = 36  < 39 = p1 x2 ,
so
(2.4) x2 R x1 .
b.  Now, from (2.2) and (2.3) we have
x1 R x0 , x0 R x2 ;
but from (2.4), something utterly revolting occurs:
x2 R x1 .
So the revealed preference  R  is   intransitivity . 

3. (MWG, 2.F.7)
Proof.   If the Walrasian demand function x( p, w) satisfies Walras’ law, then
for all p  and  w:
L
 ∂x  ( p, w)
(2.5)  p + xk ( p, w) = 0, ∀ k = 1, . . . , L ,
∂p k
=1
or
(2.6) p · D p x( p, w) + x( p, w) = 0 
and
L
 ∂x  ( p, w)
(2.7)  p = 1,
∂w
=1
or
(2.8) p · Dw x( p, w) = 1.
If the Walrasian demand function  x( p, w) is h.o.d. 0, then for all  p  and  w:
L
 ∂x  ( p, w)  ∂ x ( p, w)
(2.9)  pk  + w = 0, ∀  = 1, . . . , L ,
∂p k ∂w
k=1
or
(2.10) D p x( p, w) p + Dw x( p, w)w = 0.
 5. (JR, 2.9) 11

By the Slutsky Equation:

∂h  ( p, u) ∂x  ( p, w)  ∂ x ( p, w)
(2.11) = + xk ( p, w), ∀ , k
∂p k ∂p k ∂w
or equivalently,
(2.12) D p h( p, u) =  D p x( p, w) + Dw x( p, w)x( p, w) ,
where D p h( p, u) is equal to the matrix
 ···
s11 ( p, w) s1L ( p, w)

S ( p, w) =
 .. .. .. 
.
. .
sL1 ( p, w) · · · sLL( p, w)
with
∂x  ( p, w)  ∂ x ( p, w)
sk ( p, w) = + xk ( p, w).
∂p k ∂w
So we have
 p · S ( p, w) =  p · D p x( p, w) + p · Dw x( p, w)x( p, w) [from (2.12)]
(2.13) = p · D p x( p, w) + x( p, w) [from (2.8)]

= 0 , [from (2.6)]
and
S ( p, w) p = D p x( p, w) p + Dw x( p, w)x( p, w) p [from (2.12)]
(2.14) = D p x( p, w) p + Dw x( p, w)w [from Walras’ law]
= 0. [from (2.10)]


4. (JR, 1.62)
Solution.   Since
 p · S ( p, w) = 0,
S ( p, w) p = 0
and by symmetric, we have
a =  − 8, b = 2, p = 32.


5. (JR, 2.9)
Proof. a. In the case of two goods, equation (2.13) can be written as
 
 p · S ( p, w) =  p1 p2
s11 ( p, w) s12 ( p, w)
s21 ( p, w) s22 ( p, w)

(2.15)  
=  p1 s11 ( p, w) + p2 s21 ( p, w) p1 s12 ( p, w) + p2 s22 ( p, w)
 
= 0 0 .
From equation (2.15) we have
 p1
s21 ( p, w) =  − s11 ( p, w),
 p2
12 2. SOLUTION FOR EXERCISE II

and
 p2
s12 ( p, w) =  − s22 ( p, w).
 p1
And the equation (2.14) becomes

s ( p, w) s12 ( p, w)
S ( p, w) p = 11
 
 p1
s21 ( p, w) s22 ( p, w)  p2

(2.16) =
 
 p1 s11 ( p, w) + p2 s12 ( p, w)
 p1 s21 ( p, w) + p2 s22 ( p, w)

=

0
.
0
From equation (2.16) we have
 p1
s12 ( p, w) =  − s11 ( p, w),
 p2
and
 p2
s21 ( p, w) =  − s22 ( p, w).
 p1
Thus
s12 ( p, w) =  s 21 ( p, w).

 Part 2

Production and Choice under

Uncertainty
 CHAPTER 3

Solution for Exercise III

1. (JR, 3.36)
Proof.  We can get some intuition from the following figure 1.
x2

tx02 B
x02 y1
A
y0

O x01 tx01 x1

Figure 1.1.   Homothetic production function

Now we give the regular proof. The cost function for all   homothetic production 
 function   can be written
(3.1) c(w, y) =  h(y)φ(w),
where φ(w) is linear homogeneous. 1
Then using the  Shephard’s Lemma 2, we have
∂c(w, y)
xi (w, y) = = h(y)φi (w),
∂w i
∂c(w, y)
xj (w, y) = = h(y)φj (w).
∂w j
Hence
xi (w, y) φi ( w )
(3.2) =  ψ(w),
xj (w, y) φj (w)
that is,
∂ (xi (w, y)/xj (w, y))
= 0.
∂y
1Any cost function is linear homogeneous in the factor prices.
2JR, Theorem 3.2, p.129

15
16 3. SOLUTION FOR EXERCISE III

2. (JR, 3.39)
Proof. a:  If the production function is CRS, then
n

c(w, y) = wi xi (w, y)
i=1
n
= λf i xi (w, y) [ From the F.O.C. ]
i=1
 n
= λ xi (w, y) f i
i=1
= λy [ From the Euler Equation ]
∂c(w, y)
= y [ By the interpretation of  λ ]
∂y
or
∂c(w, y) ∂y
(3.3) = .
c(w, y) y
Solve the  partial differential equation  (3.3), we get
(3.4) c(w, y) =  yφ(w),
where φ(w) is a function of factor prices w  only.
b: If  c(w, y) = yφ(w), then from the F.O.C. of the cost minimization problem
that for  ∀  x(w, y) > 0
(3.5) wi  =  λf i ,
and
∂c(w, y)
(3.6) λ =
.
∂y
Combine equation (3.5) and (3.6), we obtain
wi  = φ(w)f i ,
or
wi xi
xi f i  = .
φ(w)
Hence, for all  x i (w, y)
n n
  wi xi
xi f i  =
i=1 i=1
φ(w)
n
1 
= wi xi
φ(w) i=1
c(w, y)
=
φ(w)
yφ(w)
=
φ(w)
= f (x).
 3. (THE ALLAIS PARADOX) 17

It follows from the  Euler’s Theorem 3 that

(3.7) f (tx) =  tf (x),
that is, the production function is CRS.


3. (The Allais Paradox)

Proof.   Defining
X   =  {x1 , x2 , x3 }  =  {$ 0 ; $ 1, 000, 000;$5, 000, 000},
these four gambles are seen form a parallelogram in the ( p1 , p3 ) triangle, as in the
following figure.
1 1

 p3  p3

g2 g3 g2 g3

g1  p1 g4 1 g1  p1 g4 1

(a) (b)

Figure 3.1.   Indifference curves and the Allais Paradox

The   Independence Axiom  is in fact equivalent to linearity in the probabilities.

IA implies that:
(1)  Indifference curves are straight lines: if,  ∀  g, g  ∈ G , we have
g  ∼  g 
implies
αg + (1 − α)g = g  ∼  αg  + (1 − α)g, ∀  α  ∈  (0, 1).
(2)  Indifference curves are parallel lines: if,  ∀  g, g  , g  ∈ G , we have
g  ∼  g  ,
the IA implies that
αg + (1 − α)g  ∼  αg  + (1 − α)g  , ∀  α  ∈  (0, 1).
3Euler’s Theorem: f (x) is homogeneous of degree  r  i ff 
n

rf (x) = xi f i .
i=1
18 3. SOLUTION FOR EXERCISE III

Now consider the conditions in the   Allais Paradox . A preference for g1 in

the first pair gambles would indicate that the individual’s indifference curves were
relatively steep, and hence a preference for  g 4  in the second pair. 4
If, on the contrary,  g 1  is preferred in the first pair, and  g 3  in the second, which
implies that indifference curves are   parallel  but rather  fan out , as in figure (b).
Now we turn to do the job follows  another way. We can rewrite the gambles
as
g1  = (0.10 ◦ 1, 0.01 ◦ 1, 0.89 ◦ 1);
g2  = (0.10 ◦ 5, 0.01 ◦ 0, 0.89 ◦ 1);
g3  = (0.10 ◦ 5, 0.01 ◦ 0, 0.89 ◦ 0);
g4  = (0.10 ◦ 1, 0.01 ◦ 1, 0.89 ◦ 0).
Consider the following three gambles
g5  = Get 1 with certainty;

g6  =

0.10
 ◦ 5,
0.01
 ◦ 0 ;

0.11 0.11
g7  = Get 0 with certainty .
By the  Completeness Axiom , we know either g 5   g6 or g 6   g5 .
(1) If g5   g6 :  by the IA, we have
0.11g5  + 0.89g5   0.11g6  + 0.89g5 ,
or
g1   g2 ;
and
0.11g5  + 0.89g7   0.11g6  + 0.89g7 ,
or
g4   g3 .
(2) If g6   g5 :  we can do the job with the same logic as part (1), and get
g2   g1 ,
and
g3   g4 .


4. (The Simplex)
Solution.   The slope:  Keeping the level of v.N.M utility constant
 p1 u(x1 ) + p2 u(x2 ) + (1 − p1  −  p2 )u(x3 ) = Const.,
and varying p 1 and p 2  alone, one has, locally,
 dp 2  dp2
u(x1 ) + u(x2 ) + ( −1 − )u(x3 ) = 0,
dp1 dp1
or
dp2 u(x3 ) − u(x1 )
= ≥  0,
dp1 u(x2 ) − u(x3 )
4In the alternative case of relatively flat indifference curves, the gambles g and  g  would be
2 3
preferred.
 5. (INDEPENDENCE AXIOM) 19

since x 2  > x3  > x1 .

Direction where the utility increasing:   Since upward movements in the
triangle increase p 2  at the expense of  p 3  (i.e. shift probability from the outcome  x 3
up to x 2 ) and leftward movements reduce  p 1  to the benefit of  p 3  (shift probability
from x1 to x3 ), these movements (and more generally, all northwest movements)
lead to stochastically dominating gambles and would accordingly be preferred.

 p2

 p1 1
Figure 4.1.   Expected utility indifference curves in the sim-
plex diagram

5. (Independence Axiom)
Proof. Let us assume without loss of generality that the elements of  A    have
been indexed so that
a1   a2   · · ·  an .
Now, let g k , 0  ≤  k  ≤  n, be the gamble that yields outcome  k  with probability
one:
gk  = (0 ◦ a1 , 0 ◦ a2 , . . . , 1 ◦ ak , . . . , 0 ◦ an ).
Then
a1   gk   an ,
since all of them can be identified with sure outcomes.
Let
g = ( p1 ◦ a1 , . . . , pn ◦ an )
be any gamble in G , then
n
g =  pk gk .
k=1
If there is only one k, s.th. pk  = 1, that is, pj = 0, ∀  j  = n, there is nothing to
prove. So let
N    = # {(1, . . . , n) : pk  
= 0, 0  ≤  k  ≤  n } > 1
20 3. SOLUTION FOR EXERCISE III

and suppose the proposition

N   −1

a1  
  pk gk  an
k=1

is true for N    − 1. By the definition of a   compound gamble ,

N   N   −1
   pk
 pk gk  = (1 − pN   ) gk  + pN   g N   .
1 − pN  
k=1 k=1

By the  induction hypothesis ,

N   −1
  pk
a1   (1 − pN   ) gk   an .
1 − pN  
k=1

Hence, by the  independence axiom , we have

N   −1
  pk
(1 − pN   )a1 + pN   g N    (1 − pN   ) gk + pN   g N    (1 − pN   )an + pN   g N   .
1 − pN  
k=1

Applying the axiom once again, we obtain

a1  = (1 − pN   )a1  + pN   a 1   (1 − pN   )a1  + pN   g N   ;
(1 − pN   )an + pN   g N    (1 − pN   )an + pN   a n  = an .
Hence, by the transitivity,
a1   g   an , ∀  g  ∈ G  .


 CHAPTER 4

Solution for Exercise IV

1. (MWG, 6.C.16)
Solution.   The maximum amount the person is willing to buy the
gamble:

(4.1) 0.5u(w − Rb − y) + 0.5u(w − Rb  + x) =  u(w),

where R b  is the maximal buying price.

u(·)

A
u(w)
B

Rb Rb

w−y w w + x w

Figure 1.1. The maximum amount the person is willing to buy

the gamble

The minimum amount the person is willing to sell the gamble:

(4.2) 0.5u(w − y) + 0.5u(w + x) =  u(w + Rs ),

where R s  is the minimal selling price.

In general, these two prices are different. However, if  u(·) is  CARA, then they
are the same. In face, equation (4.1) and (4.2) can be restated as

(4.3) CE w−R = w,

b

(4.4) CE w  =  w + Rs ,

21
22 4. SOLUTION FOR EXERCISE IV

u(·)

E [u(·)] C 
u(w) D

Rs

w−y w w + Rs w + x w

Figure 1.2.  The minimum amount the person is willing to sell

the gamble

where C E w−R and C E w are   certainty equivalence  for (4.1) and (4.2), respectively.
b

The CARA implies that1

(w − Rb ) − CE w−R = w − CE w ,
b

thus
Rb  =  Rs .


2. (MWG, 6.C.20)
Proof.
u(CE ) = 0.5u(x + ε) + 0.5u(x − ε),
where C E   is the  Certainty Equivalent . Hence

∂C E 
(4.5) u (CE ) = 0.5u (x + ε) − 0.5u (x − ε),
∂ε
2
∂ 2 CE 
(4.6) 
u (CE )
 
∂C E 
+ u (CE ) = 0.5u (x + ε) + 0.5u (x − ε).
∂ε ∂ε 2
Thus
∂CE  2
0.5u (x + ε) + 0.5u (x − ε) − u (CE )
∂ 2 CE 
 
∂ε
lim = lim
ε↓0 ∂ε 2 ε↓0 u (CE )
u (x)
= 
u (x)
=  − R a (x),
1See MWG (1995) Section 6.C.
 3. (BETWEENNESS AXIOM) 23

since
∂C E  0.5u (x + ε) − 0.5u (x − ε)
lim = lim
ε↓0 ∂ε ε↓0 u (CE )
= 0,
and
lim u(CE ) =  u(x).
ε↓0

3. (Betweenness Axiom)
Solution. The   Betweenness Axiom 2 only requires that indifference sets be
convex , i.e., if an individual is indifferent between two lotteries, then any probability
mixture of these two is equally good: if  g  ∼  g  , then
λg + (1 − λ)g  ∼  g, ∀ λ  ∈  [0, 1].
Essentially, the betweennss axiom is a   substantially weaker  version of the contro-
versial independence axiom. 3
The axioms 1-4, 6, and the betweenness axiom means that the indifference
curves are  straight lines  can be established in the same way as in Chapter 3, exercise
(3). Note that we do not use the independence axiom in that exercise, in fact,
betweenness axiom is suffices.
Note also that these straight indifference curves   need not  be parallel, because
the betweenness axiom imposes restrictions only on straight indifference curves and
nothing on the relative positions of different indifference curves.

 p2

 p1 1
Figure 3.1. Betweenness axiom means the indifference curves
are straight lines, but need not be parallel.


2See Dekel, E. (1986) for further discussion.
3See MWG Exercise 6.B.1A .
24 4. SOLUTION FOR EXERCISE IV

4. (Quadratic v.N-M Utility Function)

a.
Solution.   The restrictions are
u (w) > 0, u (w) < 0,
so
b >  − 4cw̄;
c <  0.

b.
Solution.
4c
 R a (w) =  − ,
b + 4cw
so
∂  R a (w) 16c2
=
∂w (b + 4cw)2
> 0.
This means that the quadric utility functions are  unsatisfactory . Not only do
they imply that utility reaches a maximum, they also entail that the absolute degree
of risk aversion is  increasing   in wealth, approaching infinity as utility approaches
its maximum. Consequently, one is led to the absurd result that the willingness to
gamble for a bet of fixed size should decrease as wealth is increased. 
c.
Proof.  Before going through the proof, it is worthwhile to consider the intu-
ition of the representation. The expected utility hypothesis suggests that prefer-
ences toward gambles can be represented by the   expected value  of a v.N-M utility
function  
E   u(w) ,
where w is a random variable that represents the income from an uncertain gamble.
Expected utility in general depends on the form of the function  u(·) and on the
distribution of  w. Suppose the distribution of  w  can be completely characterized
by a vector of parameters α. In particular, let w   be distributed on the real line
with a P.D.F. f (w, α). Then4
   
E   u(w)  = u(w)f (w, α) dw.

The integral on the right-hand side of this equation is a function of  .5 If we let
α

this integral be represented by

u(α),
then  
u(α) = E   u(w)
is a valid representation of preferences.
4From this subsection through the end of the chapter, we focus on   continuous   monetary
variable for convenience.
5It is  not  a function of  w  since  w  is just the variable of integration.
 4. (QUADRATIC V.N-M UTILITY FUNCTION) 25

Many problems in the economics of uncertainty are related to the trade-off 

between the average level of income and its degree of riskiness. Since the   mean  is
a summary measure of average and the  variance  is a summary measure of risk, it
will be particularly convenient to represent preferences by a function of the mean
and variance of the income distribution.  Unfortunately , this is not always possible,
because in general the mean and variance do not   completely   determine the distri-
bution of a random variable. There are many income streams that have the same
mean and variance but different probability distributions. The expected utility
associated with these income streams are different. Although  u(α) is a valid repre-
sentation of preferences, the vector α generally contains more than two parameters.
Thus a utility function that depends only on mean and variance can be at best be
viewed as an approximation to expected utility.
There are some special cases, however, when a function involving only the mean
and variance of the income distribution can be used to represent preferences. The
quadratic utility function  in this exercise is a such example.

u(w)  = a + bE   w + 2cE   w2

 
E  
  
     2 
= a + bE   w + 2c E   w + 2cVar w .

6
d.

Solution.   We prove this proposition by an indifference curve in the   mean-

variance   plane.7 To establish this result, consider two gambles g1 , g2 ∈ G    such
that
g1  ∼  g2 .

Then, the individual must be indifferent between  g 1 , g 2 , and a compound gamble

gq  = (q  ◦ g1 , (1 − q ) ◦ g2 ), q  ∈  [0, 1]

where q  denotes the probability of obtaining g1 , and consequently (1  −  q ) is the

probability of obtaining  g 2 .
gq ∼  g1  ∼  g 2 .

Letting µ i and σ i2 denote the mean and variance, respectively, of the distribu-
tions corresponding to the gambles (i = 1, 2, and  q ).

6Markowitz (1959) demonstrated that if the ordering of alternatives is to satisfy the v.N-M
axioms of rational behavior, only a quadratic utility function is consistent with an ordinal expected
utility function that depends solely on the mean and variance of the return. Consequently,   even 
if  the return for each alternative has a normal distribution, the mean-variance framework cannot
be used to rank alternatives consistently with the v.N-M axioms unless a quadratic v.N-M utility
function is specified.
7Our proof gives here follows Baron (1977).
26 4. SOLUTION FOR EXERCISE IV

V (µ, σ 2 ) =  β 

µ2 g2

µq gq
g1
µ1

σ12 σq2 σ22 σ2

Figure 4.1.   An indifference curve in the mean-variance plane.

Note that distribution functions preserve the  linear structure  of gambles (as do
P.D.F.’s),8 so

µq =
  w d qF 1 (w) + (1 − q )F 2 (w)

(4.7)
= q 
  w dF 1 (w) + (1 − q )
  w dF 2 (w)
= qµ 1  + (1 − q )µ2 .

2
σq =
   2

w − qµ 1  − (1 − q )µ2 d qF 1 (w) + (1 − q )F 2 (w)

(4.8) =
   2

q (w − µ1 ) + (1 − q )(w − µ2 ) d qF 1 (w) + (1 − q )F 2 (w)

= qσ 12 + (1 − q )σ22 + q (1 − q )(µ1 − µ2 )2 .
Solving for q  from (4.7) yields
µq  − µ2
(4.9) q  = .
µ1  − µ2
Substituting (4.9) into (4.8) yields
µ2 µ1 (µ1 − µ2 ) + µ2 σ12 − µ1 σ22
µq  −
 µ1  − µ2
 µ2q  + σq2  =
 ,
µ1 − µ22  + σ12 − σ22
2 µ21 − µ22  + σ12 − σ22
or
µq  − α µ2q  + σq2  = k,
(4.10)
 
where
µ1 − µ2
α = ,
µ21  −
µ22  + σ12 − σ22
µ2 µ1 (µ1  − µ2 ) + µ2 σ12 − µ1 σ22
k = .
µ21  − µ22  + σ12 − σ22
8See MWG (1995) p.183.