P-SET 2

Tommaso Ferracina 3199684, Ricardo Crespi, George Morris

1. Consider a consumer whose preference is represented by the utility function u : R2++ → R

defined by

1 1
u (c1 , c2 ) = − − .
c1 c2

Assume prices are strictly positive, p1 , p2 > 0.

(a) Does this preference satisfy strong monotonicity?

By definition a preference is strongly monotone if it satisfies:

a) x >> y =⇒ x ≻ y

b) x >> y =⇒ x ≿ y

Consider c = (c1 , c2 ) and c′ = (c′1 , c′2 ):

c >> c′ =⇒ c1 > c′1 and c2 > c′2

u(c) = − c11 − 1
c2
and u(c’) = − c1′ − 1
c′2
1

u(c) - u(c’) = 1
c′1
− 1
c1
+ 1
c′2
− 1
c2

Since both [ c1′ − c11 ] and [ c1′ − c12 ] are ¿ 0. u(c) - u(c’) ¿ 0 which implies u(c) ¿ u(c’) which
1 2

implies c ¿ c’.

c ≥ c′ =⇒ c1 ≥ c′1 and c2 ≥ c′2

u(c) - u(c’) = 1
c′1
− 1
c1
+ 1
c′2
− 1
c2
=⇒ c ≿ c′ By the same argument since two components
are ≥ 0, u(c) - u(c’) ≥ 0 which implies u(c) ≥ u(c’) which implies c ≥ 0 c’.

Hence, the preference is strictly monotone.

(b) Calculate the demand functions. Is there any inferior good?

M Uc1 = ∂u(c1 ,c2 )

∂c1
= 1
c21

M Uc2 = ∂u(c1 ,c2 )

∂c2
= 1
c22
M Uc1
p1
= M Uc2
p2
=⇒ 1
p1 c21
= 1
p2 c22
=⇒ p1 c21 = p2 c22

Naming the budget set p1 c1 + p2 c2 = w1 , we derive the demand functions for the two
goods.
√
For d1 : We have c2 = , substituting this into the budget set we get c1 (p1 + p1 p2 ,
q
p1
p2

therefore:
 d1 (p, w) = w
√
p1 + p1 p2

Similarly, for d1 : We have c1 = , substituting this into the budget set we get c2 (p2 +
q
p2
p1
√
p1 p2 , therefore:

d2 (p, w) = w
√
p2 + p1 p2

To check if they are normal or inferior goods we can check the derivative of the demand
function in regards to the wealth.
∂d1
∂w
= 1
√
p1 + p1 p2
> 0 (because we assumed p1 , p2 > 0)
∂d2
∂w
= 1
√
p2 + p1 p2
> 0 (because we assumed p1 , p2 > 0)

Since both the derivatives are positive, both goods are normal.

(c) Calculate the indirect utility function.

The demand functions can be plugged into u(c) to get the indirect utility function, defined
as: v(p, w) = max u(c)
1 1
v(p, w) = u(d1 (p, w), d2 (p, w) = − w
w − w
w
√ √
p1 + p1 p2 p2 + p1 p2

√ √ √
−p1 − p2 − 2 p1 p2 ( p1 + p2 ) 2
= =−
w w
(d) Are two goods substitutes, complements or independent?

We have to compute the cross-price substitution term s1,2 (p, w):

∂d1 (p, w) ∂d1 (p, w)

s1,2 (p, w) = + d2 (p, w)
∂p2 ∂w
! ! !
∂ w ∂ w w
= √ + √ √
∂p2 p1 p2 + p1 ∂w p1 p 2 + p1 p1 p2 + p2
1
! !
p1 w w
=− √ √ 2 + √ √
2 p1 p2 p1 p 2 + p1 p1 p2 + p2 p 1 p2 + p 2
√ √  √ 
2w p1 p2 p1 p2 + p1 − p1 w p1 p2 + p2
= √ √ 2 √ 
2 p1 p2 p3 p2 + p2 p1 p2 + p2
√ 
wp1 p1 p2 + p2
= √ √ 2 √  ⩾0
2 p1 p2 p1 p2 + p1 p1 p2 + p2

Since all quantities are non-negative. Hence, by definition, the two goods are substitutes.
2. Consider a consumer whose preference is represented by the Leontieff utility function u : Rn++ →
R defined by
c1 cn
 
u (c1 , ..., cn ) = min , ...,
a1 an
where ai > 0 for every i = 1, ..., n.

Assume prices are strictly positive, p1 , p2 > 0.

(a) Calculate the demand functions and the indirect utility function.

Define k1 = c1
a1
, . . . , kn = cn
an
. Then the Leontieff consumer problem becomes

max min{k1 , . . . , kn } sub p1 c1 + · · · + pn cn ≤ w.

(k1 ,...,kn )

A solution k̂ must be symmetric, i.e. k̂1 , . . . , k̂n , because, if there was an asymmetric
bundle k, we would gain the same utility without spending more money at the symmetric
bundle (min(k1 , . . . , kn ), . . . , min(k1 , . . . , kn )). Thus:

c1 p1 + · · · + cn pn = w ⇒ a1 k1 p1 + · · · + an kn pn = w,

which implies
w
k̂1 = · · · = k̂n = ,
a1 p 1 + · · · + an p n
which, unzipping the notation for ki , becomes

w
ĉi = ai ∀i ∈ {1, . . . , n}.
a1 p 1 + · · · + an p n

By assumption, p1 , . . . , pn > 0, so there is no free good and

!
w w
d(p, w) = a1 , · · · , an .
a1 p 1 + · · · + an p n a1 p 1 + · · · + an p n

Thus the value function is

v(p, w) = max min{k1 , . . . , kn }

c∈B(p,w)
( )
ĉ1 ĉn
= min ,...,
a1 an
( )
a1 w an w
= min ,...,
a1 a1 p 1 + . . . + an p n an a1 p 1 + . . . + an p n
w
= .
a1 p 1 + . . . + an p n
(b) Now consider n = 2 and a1 = a2 = 1, that is

u (c1 , c2 ) = min {c1 , c2 } .

Suppose the price of good 1 is 1 and the price of good 2 increases from 1 to 2. The
consumer’s wealth is 2. Calculate the wealth effect and the substitution effect (under
Slutsky) on the demand for good 2 and illustrate in a graph.

Consider a change in price from p = (p1 , p2 ) = (1, 1) to p′ = (p1 , p′2 ) = (1, 2), with
w = 2 initially.
If ĉ = (ĉ1 , ĉ2 ) is optimal under (p, w), by Walras we have that

p · ĉ = w, i.e. p1 ĉ1 + p2 ĉ2 = w, i.e. ĉ1 + ĉ2 = 2.

ĉ is optimal, hence ĉ1 0ĉ2 , which implies that ĉ = (1, 1) by point (a).
If we get to price p′ , Ĉ is not affordable anymore, hence there will be a new optimal bundle
ĉ′ = (ĉ′1 , ĉ′2 ). In fact,

p̂′ · ĉ′ = w if and only if ĉ′1 + 2ĉ′2 = 2,

with ĉ′1 = ĉ′2 following the same reasoning above. Hence,

2 2
 
ĉ =
′
, .
3 3

Now, we introduce the Slutsky wealth adjustment to face the change in prices, so that

p′ = ĉ′′ = w′ , i.e. ĉ′′1 + 2ĉ′′2 = 3,

with ĉ′′1 = ĉ′′2 , thus

ĉ′′ = (1, 1) = ĉ.

Hence, the wealth effect on good 2 is

2 1
d2 (p′ , w) − d2 (p, w′ ) = ĉ′2 − ĉ′′2 = −1=− .
3 3
 
Indeed, ĉ′′2 = ĉ1 , ĉ2 − 1
3
.
And the substitution effect is

d2 (p′ , w′ ) − d2 (p, w) = ĉ′′2 − ĉ2 = 1 − 1 = 0.

Indeed, ĉ′′2 = ĉ2 .

 c2

w
p2

w′
p′2

w ĉ ≡ ĉ′′
p′2
ĉ′

c1
w w′
p1 p1
3. In discussing Giffen goods, we considered the utility function

c1 − 1
u (c) =
(2 − c2 )2

defined over the convex domain (1, ∞) × [0, 2) of the plane. Check that the demand function is

1
d1 (p, w) = 2 − (w − 2p2 )
p1
2
d2 (p, w) = −2 + (w − p1 )
p2

provided w ∈ (p1 + p2 , 2p2 + p1 ).

∂u(c) 1 ∂u(c) 2 (c1 − 1)

M U c1 = = M U c2 = =
∂c1 (2 − c1 )2
∂c2 (2 − c2 )3
M U c1 M U c2
=
p1 p2
1 2 (c1 − 1) p2 (2 − c2 ) 2p1 (c1 − 1)
2 = 3; 3 = ;
p1 (2 − c2 ) p2 (2 − c2 ) p1 p2 (2 − c3 ) p1 p2 (2 − c2 )3
2p1 + 2p2 − p2 c2 p2 c2
 
2p2 − p2 c2 = 2p1 c1 − 2p1 ⇒ c1 = =1+ 1− .
2p1 p1 2

Since the budget constraint is p1 c1 + p2 c2 = w, we get

!
p2 c2 p1 c2

p1 1+ 1− + c2 p 2 = w ⇔ p1 + p 2 − + c2 p 2 = w
p1 2 2
1
⇔ c2 p 2 = w · p 1 − p 2
2
Solving for c2 gives the demand function of good 2 :

2
d2 (p, w) = −2 + (w − p1 ) .
p2

Repeating the same procedure:

2p2 + 2p1 c1 + 2p1 2p1
c2 = =2− (c1 − 1) .
p2 p2
p1 c1 + 2p2 − 2p1 (c1 − 1) = w ⇔ p1 c1 + 2p2 − 2p1 c1 12p1 = w

⇔ p1 c1 = −w + 2p2 + 2p1
2p2 2w 1
⇒ d1 (p1 w) = 2 + − = 2 − (w − 2p2 )
p1 p1 p1

4. Consider the following two lotteries, l1 = {4, 0.75; 7, 0.25} and l2 = {3, 0.5; 6, 0.25; 7, 0.25}.
(a) Which one should a risk neutral decision maker choose?

Let’s compare the expected value of each lottery:

l1 → 4 · 0.75 + 7 · 0.25 = 4.75;

l2 → 3 · 0.5 + 6 · 0.25 + 7 · 0.25 = 4.75.

Since risk-neutral DMs aim to maximize their expected monetary outcome and both lot-
teries have the same expected value, a risk neutral DM would be indifferent among the
two.
√
(b) Given two rational individuals, called A and B, with vN-M utility functions uA (c) = c,
for c ≥ 0, and uB (c) = c2 , for c ≥ 0, respectively, which lottery would each individual
choose? State whether the individuals are risk averse/ risk neutral/ risk loving.

For each individual, we analyze its utility function and then compute the expected utility
(which we’ll denote by e.u.) of the two lotteries.

√
Decision maker A: uA (c) = c, c ≥ 0.

• Since its utility function is monotonically increasing and concave, A is risk-averse.

√ √
• e.u.A (l1 ) = 4 · 0.75 + 7 · 0.25 = 2.16.
√ √ √
• e.u.A (l2 ) = 3 · 0.5 + 6 · 0.25 + 7 · 0.25 = 2.14.

• Since e.u.A (l1 ) > e.u.A (l2 ), individual A would choose lottery l1 .

Decision maker B: ub (c) = c2 , c ≥ 0.

• Since its utility function is monotonically increasing and convex, B is risk-loving.

• e.u.B (l1 ) = 42 · 0.75 + 72 · 0.25 = 24.25.

• e.u.B (l2 ) = 32 · 0.5 + 62 · 0.25 + 72 · 0.25 = 25.75.

• Since e.u.B (l1 ) < e.u.B (l2 ), individual A would choose lottery l2 .

(c) Calculate the certain equivalent and risk premium of each lottery for each individual.

Certain equivalent (c.e.) is the amount of certain wealth an individual would consider
equivalent to the uncertain prospect of a lottery. Risk premium (r.p.) is the amount an
 individual is willing to pay to avoid uncertainty and receive the certain equivalent instead.
In general, we know that
r.p. = e.u. − c.e.,

where there holds

u(c.e.) = e.u.

In our case, again, we first consider individual A and then B, computing for both the risk
premium of each lottery.

Decision maker A.

• Lottery l1 :

◦ e.u.A (l1 ) = 2.16;

◦ c.e.A (l1 ) = 2.162 = 4.67;

◦ r.p.A (l1 ) = e.u.A (l1 ) − c.e.A (l1 ) = 2.16 − 4.67 = −2.15.

• Lottery l2 :

◦ e.u.A (l2 ) = 2.14;

◦ c.e.A (l2 ) = 2.142 = 4.58;

◦ r.p.A (l2 ) = e.u.A (l2 ) − c.e.A (l2 ) = 2.14 − 4.58 = −2.44.

Decision maker B.

• Lottery l1 :

◦ e.u.B (l1 ) = 24.25;

√
◦ c.e.B (l1 ) = 24.25 = 4.92;

◦ r.p.B (l1 ) = e.u.B (l1 ) − c.e.B (l1 ) = 24.25 − 4.92 = 19.33.

• Lottery l2 :

◦ e.u.B (l2 ) = 25.75;

√
◦ c.e.B (l2 ) = 25.75 = 5.07;

◦ r.p.B (l2 ) = e.u.B (l2 ) − c.e.B (l2 ) = 25.75 − 5.07 = 20.68.

5. Consider an individual with a strictly increasing and continuous vN-M utility function of money
u (·). For any fixed amount of money c and positive number k, define the probability premium
π (c, k, u) as the excess in winning probability over fair odds that makes the individual indif-
ferent between the certain outcome c and a gamble between the two outcomes c + k and c − k.
More precisely,

1 1
   
u (c) = + π (c, k, u) u (c + k) + − π (c, k, u) u (c − k) .
2 2

Show that if the individual is risk averse (i.e., u (·) is concave), then π (c, k, u) ≥ 0 for all c, k.

Assume the individual is risk averse, then the utility function u(·) is concave. Hence

1 1
u(c) = ( + π(c, k, u))u(c + k) + ( − π(c, k, u))u(c − k)
2 2
1 1
≤ u(( + π(c, k, u))(c + k) + ( − π(c, k, u))(c − k)) , by Jensen inequality.
2 2

Since u is strictly increasing, c ≤ ( 12 + π(c, k, u))(c + k) + ( 21 − π(c, k, u))(c − k).

Hence,

1 1 1 1
c ≤ c + k + π(c, k, u)c + π(c, k, u)k + c − k − π(c, k, u)c + π(c, k, u)k = c + 2π(c, k, u)k (1)
2 2 2 2

Implication:
=⇒ 2π(c, k, u)k + c ≥ c (2)

Which means:

=⇒ π(c, k, u) ≥ 0 (3)

for all c, k, with k > 0.