ECON UN3211 - INTERMEDIATE MICROECONOMICS

Columbia University - Department of Economics

Spring 2025
Problem Set 4 - Solutions

1. A consumer can buy only two goods, good 1 denoted by x1 and good 2 denoted by x2 . Her utility function
is given by u(x1 , x2 ) = min{ax1 , bx2 }, and p1 and p2 are the prices of good 1 and good 2, respectively, and
m is the consumer’s income level. Solve for the demand functions, x1 (p1 , p2 , m) and x2 (p1 , p2 , m).
Solution: This utility function is not well-behaved. The indifference curves are L-shaped and the optimal
bundle for any given (p1 , p2 , m) will be on the kink, which is given by ax1 = bx2 . Thus, we have

a
x2 = x1
b

Then using this in the budget constraint, p1 x1 + p2 x2 = m, we get

a
p 1 x1 + p 2 x1 = m
b
a
x1 [p1 + p2 ] = m
b
bp1 + ap2
x1 [ ] = m
b
bm
x1 =
bp1 + ap2

Retrieveing x2 = ab x1 we get
a bm am
x2 = =
b bp1 + ap2 bp1 + ap2
Thus, the demand functions are

bm am
x1 (p1 , p2 , m) = and x2 (p1 , p2 , m) =
bp1 + ap2 bp1 + ap2

2. A consumer can buy only two goods, good 1 denoted by x1 and good 2 denoted by x2 . Her utility function
x1 x2
is given by u(x1 , x2 ) = x1 +x2 , and p1 and p2 are the prices of good 1 and good 2, respectively, and m is the
consumer’s income level.

(a) Is this utility function well-behaved? (Hint: It is continuous and differentiable.)

Solution: Both marginal utilities are positive:

x22 x21
M U1 = >0 and M U2 = >0
(x1 + x2 )2 (x1 + x2 )2

Thus, u(x1 , x2 ) is increasing in both x1 and x2 . Using marginal utilities, we get

M U1 x2 x2
|M RS| = = 22 = ( )2
M U2 x1 x1

This is decreasing as x1 increases and x2 decreases, implying diminishing MRS. Thus this is well-
behaved.

1
 (b) Solve for the demand functions, x1 (p1 , p2 , m) and x2 (p1 , p2 , m).
Solution: Since u(·) is well-behaved, at an interior solution
q tangency should hold. Equating MRS
to price ratio gives: |M RS12 | = ( x1 ) = p2 . Then x2 = x1 pp12 . Plugging this into the budget line,
x2 2 p1

p1 x1 + p2 x2 = m, and solving for x1 we get the demand function

m
x1 (p1 , p2 , m) = √
p1 + p1 p2

Then solving for x2 , we get the demand function

m
x2 (p1 , p2 , m) = √
p2 + p1 p2

3. A consumer can buy only two goods, good 1 denoted by x1 and good 2 denoted by x2 . Her utility function
1/2 1/2
is given by u(x1 , x2 ) = [x1 + x2 ]2 , and p1 and p2 are the prices of good 1 and good 2, respectively, and
m is the consumer’s income level.

(a) Is this utility function well-behaved? (Hint: It is continuous and differentiable.)

Solution: Both marginal utilities are positive:

1/2 1/2
1/2 1/2 1 1 x +x
M U1 = 2[x1 + x2 ] 1/2
= 1 1/2 2 >0
2x x1
1

and
1/2 1/2
1/2 1/2 1 1 x +x
M U2 = 2[x1 + x2 ] 1/2
= 1 1/2 2 >0
2x x2
2

Thus, u(x1 , x2 ) is increasing in both x1 and x2 . Using marginal utilities, we get

1/2 1/2
x1 +x2
1/2
M U1 x1 x2 1/2
|M RS| = = 1/2 1/2
=( )
M U2 x1 +x2 x1
1/2
x2

This is decreasing as x1 increases and x2 decreases, implying diminishing MRS. Thus this is well-
behaved.
(b) Solve for the demand functions, x1 (p1 , p2 , m) and x2 (p1 , p2 , m).
Solution: The Lagrangian is given by

1/2 1/2
L(x1 , x2 , λ) = [x1 + x2 ]2 + λ(m − p1 x1 − p2 x2 ).

The FOCs are

1/2 1/2
∂L x +x
= 1 1/2 2 − λp1 = 0, (1)
∂x1 x1
1/2 1/2
∂L x +x
= 1 1/2 2 − λp2 = 0, (2)
∂x2 x2
∂L
= m − p1 x1 − p2 x2 = 0. (3)
∂λ

2
 Now, equations (1) and (2) together imply

1/2 1/2 1/2 1/2

x1 + x2 x1 + x2
1/2
= 1/2
p 1 x1 p 2 x2

which is simplified to be
x2 p1
= ( )2
x1 p2
or
p1 2
x2 = x1 ( ) (4)
p2
Now equations (4) and (3) together imply

p1 2
p1 x1 + p2 x1 ( ) = m
p2
p2
p 1 x1 + x1 1 = m
p2
p1
p1 x1 [1 + ] = m
p2
p2 m
x1 =
p1 (p1 + p2 )

Then, we can retrieve x2 from equaiton (4)

p1 2 p2 m p1 p1 m
x2 = x1 ( ) = ( )2 =
p2 p1 (p1 + p2 ) p2 p2 (p1 + p2 )

That is, the demand functions are given by

p2 m p1 m
x1 (p1 , p2 , m) = and x2 (p1 , p2 , m) =
p1 (p1 + p2 ) p2 (p1 + p2 )

(c) Find the marginal utility of income. What is the change in maximized utility from an increase in
income by $10 when prices are p1 = 2 and p2 = 5?
Solution: Marginal utility of income, M Um , is equal to the Lagrangian multiplier, λ. From equation
1 we have
1/2 1/2
x1 + x2
1/2
− λp1 = 0
x1
that is
1/2
x1 + x2
1/2 [ p1 (pp21m
+p2 ) ]
1/2 + [ p1 m
p2 (p1 +p2 ) ]
1/2
[ pp21 ]1/2 + [ pp12 ]1/2 [ pp21 ]1/2 [1 + p1
p2 ] p1 + p2
λ= = = = =
1/2
p1 x1 p1 [ p1 (pp21m
+p2 ) ]
1/2 p1 [ pp12 ]1/2 p1 [ pp12 ]1/2 p1 p2

Thus,
p1 + p2
M Um =
p1 p2
p1 +p2 2+5 7
When prices are p1 = 2 and p2 = 5, we have M Um = p1 p2 = 2·5 = 10 . Thus, the utility change from
an additional $10 is
7
∆u = M Um · ∆m = 10 = 7
10

3
4. A consumer can buy only two goods, good 1 denoted by x1 and good 2 denoted by x2 . Her utility function
√
is given by u(x1 , x2 ) = x1 + x2 , and p1 and p2 are the prices of good 1 and good 2, respectively, and m is
the consumer’s income level.

(a) Suppose p1 = 1. Solve for the demand functions x1 (1, p2 , m) and x2 (1, p2 , m).
Solution: The Lagrangian is given by
√
L(x1 , x2 , λ) = x1 + x2 + λ(m − x1 − p2 x2 ).

The FOCs are

∂L 1 −1
= x1 2 − λ = 0, (5)
∂x1 2
∂L
= 1 − p2 λ = 0, (6)
∂x2
∂L
= m − x1 − p2 x2 = 0. (7)
∂λ
1
Solving equation (6) yields λ = p2 , which plugged into (5) yields

1 − 12 1 p22
x1 − =0 =⇒ x1 = ,
2 p2 4

which then plugged into (7) yields

m p2
x2 = − .
p2 4
m p2
While x1 ≥ 0 is guaranteed, but we must check x2 ≥ 0 . If, in fact, x2 = p2 − 4 ≥ 0 does not hold,
we must adopt the corner solution (x1 , x2 ) = (m, 0). Therefore the demand at prices (1, p2 ) is
  m p
p22 m p2 2 m p2

 if ≥  − if ≥
 p2 4 p2 4
 
 4
 p2 4 
x1 (1, p2 , m) = and x2 (1, p2 , m) =

 m p2 
 m p2
 m

 if <

 0 if <
p2 4 p2 4

(b) Now find the demand functions for a general prices (p1 , p2 ). (Hint: For any α > 0 and prices and
income, (p1 , p2 , m), the budget set under (αp1 , αp2 , αm) is the same budget set under (p1 , p2 , m), that
is, when all prices and income change proportionately the budget set does not change. Thus, when all
prices and income change proportionately, the demand does not change either. Thus, x(p1 , p2 , m) =
x(αp1 , αp2 , αm), for any α > 0. Now set α = 1/p1 .)
Solution: Setting α = 1/p1 ,

p2 m p2 m
x1 (p1 , p2 , m) = x1 (1, , ) and x2 (p1 , p2 , m) = x2 (1, , )
p1 p1 p1 p1
p2 m
Therefore, replacing p2 with p1 and m with p1 , we get
 2  m
p2 m p2 p2 m p2

 if ≥  − if ≥
 4p21  p2 4p1 p2 4p1
 
 p2 4p1 
x1 (p1 , p2 , m) = and x2 (p1 , p2 , m) =

 m m p2

 m p2

 if <

 0 if <

p1 p2 4p1 p2 4p1

4
 (c) Find the marginal utility of income in terms of prices and income.
Solution: We know that the Lagrangian multiplier attached to the the budget constraint is the
marginal utility of income (shadow price of income, as it’s often called). From equation (6) above we
1
already know that λ = p2 , when the solution is interior, which is the case when the income is high
enough: m/p2 ≥ p2 /4p1 . In this case the utility function calculated at the optimal bundle is given by
s
p22 m p2
u(x1 (p1 , p2 , m), x2 (p1 , p2 , m)) = u(p1 , p2 , m) = 2 +( − )
4p1 p2 4p1

You can see that the partial derivative of v(p1 , p2 , m) with respect to m is 1/p2 .
m p2
However when the income is not large enough then we have a corner solution: if p2 < 4p1 then the
optimal bundle is x1 = m/p1 and x2 = 0. Thus, the utility function calculated at the optimal bundle
is given by r r
m m
u(p1 , p2 , m) = +0=
p1 p1
q
m √1
The partial derivative of p1 with respect to m is 2 mp1 . Thus we have the marginal utility of income
given as
1 m p2


 if ≥


 p2 p2 4p1
M Um (p, m) = λ(p, m) =

 1 m p2
√ if <


2 mp1 p2 4p1


5. A consumer can buy only two goods, good 1 denoted by x1 and good 2 denoted by x2 . The prices of good
1 and good 2 are p1 and p2 , respectively, and m is the consumer’s income level. For each of the following
utility functions below, for good 1, derive and draw the demand curve, x1 (p1 ), and the Engel curve, x1 (m).
1/3 2/3
(a) u(x1 , x2 ) = x1 x2
Solution: We can get the |M RS| as

−2/3 2/3
M U1 (1/3)x1 x2 1 x2
|M RS| = = 1/3 −1/3
=
M U2 (2/3)x1 x2 2 x1

p1 1 x2 p1 2p1 x1
Tangency, |M RS| = p2 , then gives 2 x1 = p2 that is x2 = p2 . Plugging this into the budget
constraint, p1 x1 + p2 x2 = m, we get, p1 x1 + p2 2pp12x1 = m, that is, p1 x1 [1 + 2] = m, that is x∗1 = m
3p1 .
1m
Thus, the demand function is x1 (p1 ) = 3 p1 (where m is given) and it is depicted below. The Engel
1m
curve is x1 (m) = 3 p1 (where p1 is given) and it is depicted below.

p1 m

Demand curve Engel curve

slope : 3p1

x1 x1

5
(b) u(x1 , x2 ) = 2x1 + 3x2
Solution: Note that this is a perfect substitute case. |M RS| is constant and equal to 2/3.
p1
If p2 > 32 , then x∗1 = 0.
p1
If p2 = 32 , then x∗1 ∈ [0, pm1 ].
p1
If p2 < 32 , then x∗1 = m
p1 . Thus, the demand curve, given p2 and m, is



 0 if p1 > 32 p2





 h i
m
x1 (p1 ) = 0, p1 if p1 = 23 p2




 m



 if p1 < 23 p2
p1
And the Engel curve, given p1 and p2 , is



 0 if p1 > 23 p2





 h i
m
x1 (m) = 0, p1 if p1 = 23 p2





 m
if p1 < 23 p2



p1

p1 m m m
Engel curve Engel curve Engel curve
3p1 > 2p2 3p1 < 2p2 3p1 = 2p2
Demand curve

2p2 /3

m/x1 = p1 slope : p1
m = p 1 x1 slope : p1
m = p 1 x1

3m/2p2 x1 x1 x1 x1

1/2
(c) u(x1 , x2 ) = x1 + 2x2
−1/2
M U x1 1/2(x1 ) 1
Solution: |M RS| is given by M U x2 = 2 = 1/2 . This is diminishing as x1 goes up. Then,
4x1
1 p1 p2 2
tangency implies 1/2 = p2 , which gives x1 = ( 4p 1
) . If m is large enough, the demand will be
4x1
p2 2
x1 = ( 4p 1
) , otherwise it will be x1 = m/p1 . Thus we have
p22 p2 2
If m ≥ 16p1 , x∗1 = ( 4p1
) .
p2
If m < 16p2 1 , x∗1 = m
p1 . Thus, the demand curve, given p2 and m, is

p22 1 p22


 16 p21 if p1 ≥ 16m

x1 (p1 ) =
 m p22
if p1 <



p1 16m

6
 And the Engel curve, given p1 and p2 , is

p22

( p2 )2 if m≥
 4p1 16p1


x1 (m) =
 m p22
if m<


p1 16p1

p1 m

Demand curve Engel curve

p22 1
x1 = 16 p21

p22
16p1
p22
16m

m
x1 = p1

16m2
p22
x1 p2 2
( 4p ) x1
1

(d) u(x1 , x2 ) = x21 + x22

Solution: The |M RS| is equal to
M U x1 2x1 x1
= =
M U x2 2x2 x2
which is increasing in x1 and decreasing in x2 . Thus, MRS is not diminishing. The solution will be a
corner solution.
If p1 > p2 , then spend all income on x2 , that is, x∗1 = 0.
If p1 = p2 , then the optimal bundle is either (x∗1 = 0, x∗2 = m
p2 ) or (x∗1 = m ∗
p1 , x2 = 0).
If p1 < p2 , then spend all income on x, that is, x∗1 = m
p1 . Thus, the demand curve, given p2 and m, is



 0 if p1 > p2





 m
x1 (p1 ) = 0 and if p1 = p2
 p1



m




 if p1 < p2
p1

And the Engel curve, given p1 and p2 , is




 0 if p1 > p2





 m
x1 (m) = 0 and if p1 = p2
 p1



m




 if p1 < p2
p1

7
 p1 m m m
Demand curve Engel curve Engel curve Engel curve
p1 > p2 p1 < p2 p1 = p2

p2

slope : p1
m/x1 = p1
slope : p1
m = p 1 x1
m = p 1 x1

m/p2 x1 x1 x1 x1

6. Suppose a consumer spends all of her income on only two goods, x and y. Her preferences over these two
goods are represented by the utility function u(x, y) = min{2x + y, x + 3y}. Let px and py denote the prices
for good x and good y respectively. Consumer has an income m.

(a) Draw the demand curve for good x, for py = 1 and m = 100.
Solution: When x > 2y, we have 2x + y > x + 3y, thus u(x, y) = x + 3y and the |M RS| = 1/3. When
x < 2y, we have 2x + y < x + 3y, thus u(x, y) = 2x + y and the |M RS| = 2. Therefore, the indifference
curves are as in the graph below.

px
Given py = 1 and m = 100, the optimal x demand depends on the price ratio, which is py = px .

If px > 2, then the price ratio is larger than the |M RS| of the upper portion of the indifference curves,
thus the budget line is steeper than the upper portion of the indifference curves. Then the optimal
bundle is a corner solution with all income spent on y. Thus, x∗ = 0 for any px > 2.

If 2 > px > 1/3, then the price ratio is between the two |M RS| values of the upper and lower portions
of the indifference curves. Then the optimal bundle is at the kink of an indifference curve. Thus,
x∗ = 2y ∗ . Inserting this into the budget line, we get px x + py y = 100, we get px x + x
2 = 100, that is,
for any px with 2 > px > 1/3, we have
200
x∗ =
1 + 2px

8
If 1/3 > px , then the price ratio is smaller than the |M RS| of the lower portion of the indifference
curves, thus the budget line is flatter than the lower portion of the indifference curves. Then the
optimal bundle is a corner solution with all income spent on x. Thus, for any 1/3 > px we have

100
x∗ =
px

Finally let’s consider the prices 2 and 1/3:

If px = 2, then the price ratio is equal to |M RS| of the upper portion of the indifference curves, thus
the budget line overlaps with the steeper/upper portion of the indifference curves. Then any bundle
on the upper portion of the corresponding indifference curve will be optimal. Thus, for px = 2, optimal
x∗ ranges between 0 and the value at the kink, 200
1+2px at px = 2, which makes the latter 40. Thus,
x∗ ∈ [0, 40] for px = 2.

If px = 1/3, then the price ratio is equal to |M RS| of the lower portion of the indifference curves, thus
the budget line overlaps with the flatter/lower portion of the indifference curves. Then any bundle on
the lower portion of the corresponding indifference curve will be optimal. Thus, for px = 1/3, optimal
x∗ ranges between the value at the kink, 200
1+2px and the corner solution at 100
px . Thus, at px = 1/3,
optimal x∗ ranges between 200
1+ 23
= 120 and 100
1 = 300. Thus, x∗ ∈ [120, 300] for px = 1/3.
3
See the graph below for each of these prices and price ranges.

Now, if we put all these together to find the demand curve x(px ), we get the following graph.

9
(b) Draw the Engel curve for good x, for px = 2 and py = 3.
Solution: When prices are px = 2 and py = 3, the price ratio is 2/3 which is between the two
|M RS| values of the upper and lower portions of the indifference curves. Then the optimal bundle is
at the kink of an indifference curve. Thus, x = 2y. Inserting this and the prices into the budget line,
px x + py y = m, we get 2x + 3y = m, that is, 4y + 3y = m, which gives y ∗ (m) = m
7. Thus, x∗ (m) = 2m
7 ,
which gives the Engel curve in the graph below.

(c) Draw the Engel curve for good x, for px = 1 and py = 3.

Solution: When prices are px = 1 and py = 3, the price ratio is 1/3 which overlaps with the
flatter/lower portion of the indifference curves. Then, for a given m, any bundle on the lower portion
of the corresponding indifference curve will be optimal, x ranges from the the optimal x at the kink
to the corner solution where all income spent on x. To find the value at the kink, again insert x = 2y
into the budget line at prices px = 1 and py = 3, and we get x + 3y = m, that is 2y + 3y = m, that is
y ∗ (m) = m
5 and x∗ (m) = 2m
5 . The corner solution is given by x∗ (m) = m
px = m. Thus, for any given
m, we get x∗ ∈ [ 2m
5 , m]. This gives the Engel curve in the graph below.

10
7. Determine in the following scenarios whether the good in question is a normal good for sure, an inferior
good for sure, or whether one cannot tell from the information given.

(a) When Justin was a graduate student with very low income in NYC, he always ate burritos from a food
truck for lunch. Now Justin has graduated and has a well-paying job in Toronto, where burritos from
food trucks are cheaper than they were in NYC. Justin eats far fewer burritos now even though he
has just as much time for lunch, his preferences haven’t changed, the burritos are just as good and the
lunch alternatives are more or less the same. For Justin, what kind of good are food truck burritos?
Solution: Burritos have gotten cheaper relative to other lunch alternatives, and so the substitution
effect alone must be positive. However the total effect is observed to be negative (since Justin now eats
fewer burritos), meaning the income effect must be negative. Since Justin now has a higher income
(in two senses!–Justin has a higher income in nominal terms and also has a higher real income due to
the decrease in the price of burritos), the income effect being negative means that the burritos must
be an inferior good.
(b) In his grad student days, Justin hardly ever bought himself a lemonade, let alone buying lemonade
for others. Now, when he comes back to visit NYC, it’s quite possible to see him buy entire rounds
of lemonades, even for strangers. The price of lemonade in NYC has not changed, nor have Justin’s
preferences. For Justin, what kind of good is lemonade?
Solution: Essentially nothing has changed here except for Justin’s income. Since an increase in income
has led to an increase in his consumption of lemonade, lemonade must be a normal good.
(c) Since Justin got his job in Toronto, the Canadian dollar (the currency in which Justin is paid his
salary) has been becoming increasingly more valuable relative to the U.S. dollar. Because Justin visits
the U.S. quite often, it is as easy for him to shop for clothes in either country. Lately, he has been
doing more and more of his clothes shopping in the U.S. The price tags have not changed in either
country, and the clothes Justin buys are the same kind in both countries. For Justin, what kind of
good are clothes?
Solution: The question is asking about clothes in general, not clothes in one particular country.
Therefore, all we are really observing is that Justin is choosing to buy clothes from the cheaper place
(the U.S.) rather than the more expensive one (Canada). This tells us nothing whatsoever about
income effect. Rather, it tells us Justin is good at understanding the exchange rate and wants to buy
identical things for less rather than more. While this leads us to believe that Justin is normal, we
cannot say anything about whether clothes are normal or not.

11