ECON 357: Extra Practice Problems - Preferences

and Utility

Erik Lillethun
Colgate University

Note: For problems marked (Varian), the source is

Varian, H. R. (1992). Microeconomic Analysis (Vol. 3). New York: Norton.

For problems marked (MWG), the source is

Mas-Colell, A., Whinston, M. D., & Green, J. R. (1995). Microeconomic Theory (Vol.
1). New York: Oxford University Press.

1. For each of the following preference relations on X = R2+ , either prove or come up
with a counterexample for completeness, transitivity, monotonicity, convexity, and
continuity:

(a) [x ≿ y] ⇔ [min{x1 , x2 } ≥ min{y1 , y2 }]

Solution. Completeness: True. Since it is a comparison between the same
function (min function) of two numbers, the inequality must hold one way or
the other.
Transitivity: True. Again, real numbers with ≥ are transitive, and this is
based on comparing the same real-valued function applied to the two bundles.
Monotonicity: True, but only weak monotonicity. x ≥ y ⇔ x1 ≥ y1 and
x2 ≥ y2 . Then, min{x1 , x2 } ≥ min{y1 , y2 }, so x ≿ y. If x > y, which is
equivalent to x1 > y1 and x2 > y2 , then min{x1 , x2 } > min{y1 , y2 }, so x ≻ y.
Convexity: True. Let x ≿ y, which is equivalent to min{x1 , x2 } ≥ min{y1 , y2 }.
Then for any α ∈ (0, 1),

min{αx1 + (1 − α)y1 , αx2 + (1 − α)y2 } ≥ α min{x1 , x2 } + (1 − α) min{y1 , y2 }

≥ min{y1 , y2 }

0
 The first inequality above is true, because minimizing the two terms individu-
ally is minimizing over more possible combinations. For example, on the right
hand side, the min’s could select x1 and y1 or x2 and y2 , which gives the same
value as on the left hand side. But they could select x1 and y2 or x2 and y1 ,
which was not possible on the left hand side. The second inequality follows
from x ≿ y and the fact that a weighted average of two numbers is always at
least as big as the lower of the two numbers.
Continuity: True. This comes pretty readily from the following fact about
sequences:
Theorem 1. Suppose that xn ≥ y n for all n ∈ {1, 2, . . .} and that xn → x and
y n → y. Then, x ≥ y.

In other words, the ≥ relation on R satisfies the continuity assumption.

Note that min{xn1 , xn2 } and min{y1n , y2n } are also sequences that converge to
limits, so Theorem 1 directly implies that preference is preserved in the limit.

(b) [x ≿ y] ⇔ [x1 ≥ y1 or x2 ≥ y2 + 1]
Solution. Completeness: True. For any x, y, either x1 ≥ y1 or y1 ≥ x1 (in
other words, it follows from the completeness of ≥ on R.
Transitivity: False. Let x = (2, 2), y = (1, 4), z = (3, 2). We have x ≿ y,
because x1 = 2 ≥ 1 = y1 , and y ≿ z, because y2 = 4 ≥ 3 = z2 + 1. However,
x1 = 2 < 3 = z1 and x2 = 2 < 3 = z2 + 1, so z ≻ x.
Monotonicity: True. First, x ≥ y ⇒ x1 ≥ y1 , so x ≿ y. Then, xk > yk , ∀k
implies that y1 < x1 and y2 < x2 < x2 + 1. Thus, x ≻ y.
Convexity: False. Let x = (0, 2) and y = (1, 1). Then, x ≿ y, because
x2 = 2 ≥ 2 = y2 + 1. But for α = 12 , z ≡ αx + (1 − α)y = ( 12 , 32 ). Since
z1 = 21 < y1 and z2 = 32 < 2 = y2 + 1, we have y ≻ z.
Continuity: True. This comes again from Theorem 1. If xn1 ≥ y1n for all n,
it follows that x1 ≥ y1 , so x ≿ y. Or, if xn2 ≥ y2n + 1 for all n, then x2 ≥ y2 + 1,
and x ≿ y. There is an added complication if the sequence of bundles were
to bounce back and forth between satisfying the two “or” conditions, but don’t
worry about that detail. It still works out, and the relation is continuous.

(c) [x ≿ y] ⇔ [||x − (2, 3)|| ≤ ||y − (2, 3)||]

Solution. Completeness: True. For any x, y, ||x − (2, 3)|| and ||y − (2, 3)||
are real numbers. ≥ is complete on the reals, so either x ≿ y or y ≿ x.

1
 Transitivity: True. For any x, y, z with x ≿ y and y ≿ z, we have

||x − (2, 3)|| ≤ ||y − (2, 3)|| ≤ ||z − (2, 3)||,

so ||x − (2, 3)|| ≤ ||z − (2, 3)|| and therefore x ≿ z. In other words, it follows
from the transitivity of ≥ on R.
Monotonicity: False. Let x = (3, 3) and y = (2, 3). x ≥ y, but ||x−(2, 3)|| =
1 > 0 = ||y − (2, 3)||, so y ≻ x.
Convexity: True. Suppose x ≿ y. For any α ∈ [0, 1], let z ≡ αx + (1 − α)y.
Then,
p
(z1 − 2)2 + (z2 − 3)2
||z − (2, 3)|| =
p
= [α(x1 − 2) + (1 − α)(y1 − 2)]2 + [α(x2 − 3) + (1 − α)(y2 − 3)]2

We need to show that

p p
[α(x1 − 2) + (1 − α)(y1 − 2)]2 + [α(x2 − 3) + (1 − α)(y2 − 3)]2 ≤ (y1 − 2)2 + (y2 − 3)2
⇔ [α(x1 − 2) + (1 − α)(y1 − 2)]2 + [α(x2 − 3) + (1 − α)(y2 − 3)]2 ≤ (y1 − 2)2 + (y2 − 3)2
⇐ α(x1 − 2)2 + (1 − α)(y1 − 2)2 + α(x2 − 3)2 + (1 − α)(y2 − 3)2 ≤ (y1 − 2)2 + (y2 − 3)2 ,

where the last line follows from f (x) = (x−2)2 being a convex function. Then,
rearranging and dividing by α we get

(x1 − 2)2 + (x2 − 3)2 ≤ (y1 − 2)2 + (y2 − 3)2 ,

which follows from ||x − (2, 3)|| ≤ ||y − (2, 3)||. Whew! That one was a lot of
work (and a bit tedious), so I certainly would not ask you that on a quiz or
exam. But, it’s good practice for understanding the concepts.
Continuity: True. For any sequences xn → x and y n → y with xn ≿ y n
for all n, we have that ||xn − (2, 3)|| ≤ ||y n − (2, 3)|| for all n. The left and
right hand sides of the inequality are sequences of real numbers. Since ≥ is
continuous on the reals, the limits satisfy ||x − (2, 3)|| ≤ ||y − (2, 3)||, so x ≿ y.

(d) [x ≿ y] ⇔ [||x|| > ||y||]

Solution. Completeness: False. It even violates the more basic reflexivity
assumption. Take any bundle x. Since ||x|| ≤ ||x|| we cannot conclude that
x ≿ x. You also get a counterexample for any x, y equidistant from the origin,

2
 like x = (1, 0), y = (0, 1).
Transitivity: True. If x ≿ y and y ≿ z, then ||x|| > ||y|| > ||z||, so
||x|| > ||z|| and x ≿ z.
p p
Monotonicity: True. x ≥ y and x ̸= y implies that x21 + x22 > y12 + y22
and thus x ≿ y and ¬(y ≿ x), so x ≻ y. This comes from the fact that
√
f (a, b) = a2 + b2 with a, b ≥ 0 is a strictly increasing function of both a and
b (check the partial derivatives). In fact, this also implies that we have strong
monotonicity (though you were not asked that).
Convexity: False. Let x = (1.02, 0) and y = (0, 1). ||x|| > ||y|| q = 1, so
x ≿ y. Then, if α = 2 , z ≡ αx + (1 − α)y = (0.51, 0.5). But ||z|| ≈ 12 < 1,
1

so y ≻ z.
Continuity: False. Let xn = (1 + n1 , 1) and y n = (1, 1). Then, xn ≿ y n for
all n (strong monotonicity). However, xn → (1, 1) and y n → (1, 1), but we do
not have (1, 1) ≿ (1, 1) (reflexivity fails). Note that we cannot use that special
theorem about limits of sequences of real numbers, because that is only true for
≥, not >.

(e) [x ≿ y] ⇔ [x1 + 2x2 ] ≥ [2y1 + y2 ]

Solution. Completeness: False. It fails even the more basic reflexivity.
Consider x = (1, 0). We do not have x ≿ x, because 1 + 2(0) < 2(1) + 0.
Transitivity: False. Consider the example x = (1, 2), y = (0, 4), z = (2, 3).
x ≿ y because 1 + 2(2) = 5 ≥ 4 = 2(0) + 4. y ≿ z, because 0 + 2(4) = 8 ≥ 7 =
2(2) + 3. However, x is not weakly preferred to z, because 1 + 2(2) = 5 < 7 =
2(2) + 3.
Monotonicity: False. Consider x = (5, 1) and y = (4, 0). x > y but
5 + 2(1) = 7 < 8 = 2(4) + 0, so x is not weakly preferred to y.
Convexity: False. Consider y = (1, 0) and x = (2, 0). x ≿ y, because
2 + 2(0) ≥ 2(1) + 0. However, for α = 21 , αx + (1 − α)y = (1.5, 0). We do
not have αx + (1 − α)y ≿ y, because 1.5 + 2(0) < 2(1) + 0. Note that this
counterexample comes from the failure of completeness, which often causes the
other assumptions to fail.
Continuity: True. Again, it comes from Theorem 1 and the fact the x1 + 2x2
and 2y1 + y2 are both continuous functions.

(f) [x ≿ y] ⇔ [x1 ≥ y1 and x2 ≥ y2 ]

Solution. Completeness: False. Counterexample x = (1, 0) and y = (0, 1).

3
 Transitivity: True. Suppose x ≿ y and y ≿ z. Then, x1 ≥ y1 and y1 ≥ z1 ,
so x1 ≥ z1 . Likewise, x2 ≥ y2 and y2 ≥ z2 , so x2 ≥ z2 . Together, this implies
that x ≿ z.
Monotonicity: True. The weak part of the definition of monotonicity is
equivalent to the ⇐ part of the definition of the preference relation. Now
suppose that x > y. This implies that it is not true that y1 ≥ x1 , so the first
condition in the “and” fails, so we do not have y ≿ x. Therefore, x ≻ y.
Convexity: True. Suppose that x ≿ y. It follows that x1 ≥ y1 and x2 ≥ y2 .
For any α ∈ (0, 1), this implies αx1 + (1 − α)y1 ≥ y1 and αx2 + (1 − α)y2 ≥ y2 ,
which is equivalent to αx + (1 − α)y ≿ y.
Continuity: True. Consider any sequences {xn }, {y n } with xn ≿ y n for
all n, and xn → x, y n → y. This implies xn1 ≥ y1n and xn2 ≥ y2n for all n.
Theorem 1 applies individual for each of these inequalities, so x1 ≥ y1 and
x2 ≥ y2 . This is equivalent to x ≿ y.

2. For each of the preference relations in the previous question, graph the upper
contour set, the lower contour set, and the indifference set for the bundle (3, 3).

Solution. All solutions have the UCS in blue, the LCS in red, and the indifference
set in solid or shaded black.

(a)

4
(b)

(c)

5
(d)

(e)

6
 (f )

3. (Varian Exercise 7.1) Consider preferences defined over the nonnegative orthant
by (x1 , x2 ) ≻ (y1 , y2 ) if x1 + x2 < y1 + y2 . Do these preferences exhibit local
nonsatiation? If these are the only two consumption goods, and the consumer faces
positive prices, will the consumer spend all of his income? Explain.

Solution. They exhibit local nonsatiation at every point except for (0, 0). Consider
any x ∈ R2+ . If xi > 0, then the bundle z ≡ (xi − ε, x−i ) for any small ε > 0 has
z1 + z2 = x1 + x2 − ε < x1 + x2 , so z ≻ x. This is not true for x = (0, 0), because
x1 + x2 = 0 < z1 + z2 for every other bundle z. They do not satisfy any form of
monotonicity, because adding to either good makes the bundle worse.
With positive prices, (0, 0) is most preferred and it is always affordable with expen-
diture 0, so the consumer will not spend all of their income.

4. Show that the following utility functions are quasi-concave:

(a) u(x) = min{x1 , . . . , xN }

7
 Solution. Consider any two bundles x, y and any α ∈ [0, 1]. Then,

u(αx + (1 − α)y) = min{αx1 + (1 − α)y1 , . . . , αxN + (1 − α)yN }

≥ min{αx1 , . . . , αxN } + min{(1 − α)y1 , . . . , (1 − α)yN }
= αu(x) + (1 − α)u(y)
≥ min{u(x), u(y)}

This actually proves that the utility function is concave (next to last line), not
just quasi-concave. The tricky part is the second step, which exploits the fact
that minimizing each term in the sum separately makes the sum smaller than
minimizing it all at once. For example, if x = (2, 6), y = (8, 4), α = 21 , then
min{αx1 +(1−α)y1 , αx2 +(1−α)y2 } = min{5, 5} = 5. But, min{αx1 , αx2 } = 1
and min{(1 − α)y1 , (1 − α)y2 } = 2, which sum to only 3.

(b) u(x) = ax1 + bx2 , where a, b > 0

Solution. Below is a general proof for any number of goods. But you could
also just show that the MRS is constant − ab .
Consider any two bundles x, y and any α ∈ [0, 1]. Then,

u(αx + (1 − α)y) = [αx1 + (1 − α)y1 ] + . . . + [αxN + (1 − α)yN ]

= α(x1 + . . . + xN ) + (1 − α)(y1 + . . . + yN )
= αu(x) + (1 − α)u(y)
≥ min{u(x), u(y)}

So the utility function is quasi-concave (and actually concave; see the next to
last line).
√ √
(c) u(x) = x1 + x2
Solution. Below is a generalqproof for any number of goods. But you could
just show that the MRS = − xx21 exhibits diminishing MRS, because as you
move down and right along an IC, x2 (numerator) falls and x1 (denominator)
rises, causing the magnitude of the MRS to shrink.

8
 Consider any two bundles x, y and any α ∈ [0, 1]. Then,
p p
u(αx + (1 − α)y) = αx1 + (1 − α)y1 + . . . + αxN + (1 − α)yN
√ √ √ √
≥ [α x1 + (1 − α) y1 ] + . . . + [α xN + (1 − α) yN ]
√ √ √ √
= α( x1 + . . . + xN ) + (1 − α)( y1 + . . . + yN )
= αu(x) + (1 − α)u(y)
≥ min{u(x), u(y)}

Again, this is concave, not just quasi-concave. Moving from the first line to
the second is a direct application of the definition of concave function, applied
to the square root function.

5. For each of the following utility functions on R2+ , find the marginal utilities to show
that they are monotonic. Then, find the marginal rate of substitution (MRS). Using
the concept of diminishing MRS, argue why they are or are not quasi-concave utility
functions.

(a) U (x1 , x2 ) = x31 x22

Solution. M U1 = 3x21 x22 ≥ 0 and M U2 = 2x31 x2 ≥ 0. If we are being picky,
both of these marginal utility are 0 when x1 = x2 = 0, but this gives U (0, 0) = 0
and for any δ > 0, U (δ, δ) > 0, so it satisfies montonicity. I will omit this
detail in future answers, and you can ignore it on exams, as well.
3x2 x2
M RS = − 2x13 x22 = − 32 xx12 . This becomes smaller in magnitude as you move
1
down and to the right (higher x1 increases the denominator, and smaller x2
shrinks the numerator), so this satisfies diminishing MRS and it is a quasi-
concave utility function.

(b) U (x1 , x2 ) = x1 + ln(x2 )

Solution. M U1 = 1 > 0, M U2 = x12 > 0, so this satisfies monotonicity.
M RS = − 11 = −x2 . This shrinks as you move down (lower x2 ) and to the
x2
right, so this is a quasi-concave utility function.

(c) U (x1 , x2 ) = x21 + 3x2

Solution. M U1 = 2x1 ≥ 0 and M U2 = 3 > 0, so this satisfies monotonicity.
M RS = − 23 x1 . This increases in magnitude as you go down and to the right
(higher x1 ), so this is not a quasi-concave utility function.
√ √
(d) U (x1 , x2 ) = 2 x1 + 3 x2

9
 −2
Solution. M U1 = √1 > 0, M U2 = 13 x2 3 > 0, so this satisfies monotonicity.
x1
2
x3
M RS = −3 √x2 1 .
This diminishes as you go down (smaller numerator) and to
the right (larger denominator), so this is a quasi-concave utility function.
√
(e) U (x1 , x2 ) = x1 + x22
Solution. M U1 = 2√1x1 > 0, M U2 = 2x2 ≥ 0, so this satisfies monotonicity.
M RS = − 2√x11 x2 . It is not totally clear from this alone, because moving down
and to the right has offsetting effects. Consider any indifference curve for
√ p √
utility level u. The equation for this is u = x1 + x22 ⇔ x2 = u − x1 .
Substituting this into the M RS yields − √ √1 √ . As x1 increases, the
2 x1 u− x1
denominator first moves up and then moves down. This gives us a violation
of diminishing MRS (sometimes moving right along and indifference curve
increases the magnitude of the slope), so this is not quasi-concave.

6. Produce a counterexample showing that the following utility functions are not
quasi-concave:

(a) u(x) = max{x1 , . . . , xN }

1
Solution. Let x = (1, 0, . . . , 0) and y = (0, 1, 0, . . . , 0) and α = 2
. Then,
u(x) = u(y) = 1, but u(αx + (1 − α)y) = 12 < 1.

(b) u(x) = x21 + . . . + x2N

Solution. The same counterexample as the previous question works here as

2
2
well. u(x) = u(y) = 1, but u(αx + (1 − α)y) = 12 + 21 = 12 < 1.

10