Consumer Theory

Rafayal Ahmed

NSU SBE

Rafayal Ahmed (NSU) Demand 1 / 20

The Consumer’s Choice Problem

▶ Consider a price-taking consumer, who has to choose a

consumption bundle containing non-negative amounts of L
goods.
▶ Remember, consumption bundles are vectors: x = (x1 , ..., xL ).
▶ Suppose the consumer faces market prices p = (p1 , ..., pL ).
▶ She has to choose a bundle x ∈ X , subject to limited budget
m. We assume X = RL+ .
▶ Throughout, we will also assume that prices are non-negative
real numbers: p ∈ RL+ .
▶ The consumer has rational and continuous preferences over
bundles, represented by the continuous utility function
u : X → R.
▶ Informally, the consumer’s decision problem is: choose the
“best” bundle x ∈ X , subject to p.x ≤ m.

Rafayal Ahmed (NSU) Demand 2 / 20

Utility Maximization Problem

▶ Because her preferences can be represented by u, the best

bundle x corresponds to highest value of u (x ).
▶ Therefore, the consumer’s utility maximization problem
(UMP) is:

max u (x )
x ∈RL+

subject to p.x ≤ m

Rafayal Ahmed (NSU) Demand 3 / 20

Analyzing the UMP

▶ The undergrad approach to studying the UMP is solving the

problem for very simple utility functions with only two goods.
▶ Expect such childish tasks in exams and problem sets.
▶ In reality, consumers choose from hundreds of consumption
goods.
▶ However, solving a UMP with 139 goods for a nasty-looking
utility function is a daunting task.
▶ In class, we will instead be focusing on learning what’s true
for any reasonable utility function.
▶ This will often mean proving results that apply to the
solution(s) of the UMP, without actually solving the problem.

Rafayal Ahmed (NSU) Demand 4 / 20

The Budget Set

▶ Notice that because of the budget constraint, the consumer

can only choose x from a limited subset of X .
▶ Namely, she has to choose some x such that p.x ≤ m.
▶ We will call the set of all affordable bundles the consumer’s
budget set.
▶ Because the set of affordable bundles depend on the price
vector and the consumer’s income, we denote the budget set
as B (p, m).
▶ Formally, B (p, m) = {x ∈ X : p.x ≤ m}.

Rafayal Ahmed (NSU) Demand 5 / 20

Budget Sets are Convex
Lemma
For any given p and m, the budget set B (p, m) is a convex set.
Proof.
Suppose x and y are two bundles in the budget set, and take arbitrary
α ∈ [0, 1]. Consider the convex combination αx + (1 − α) y . We have to show
that αx + (1 − α) y ∈ B (p, m) as well.
Now, because x ∈ B (p, m), we have p.x ≤ m, so α (p.x ) ≤ αm.
Now, because y ∈ B (p, m), we have p.y ≤ m, so (1 − α) (p.y ) ≤ (1 − α) m.
Adding the two inequalities, we get:

α (p.x ) + (1 − α) p.y ≤ αm + (1 − α) m
p. [αx ] + p. [(1 − α) y ] ≤ m
p. [αx + (1 − α) y ] ≤ m

Thus, αx + (1 − α) y ∈ B (p, m) as well.

Rafayal Ahmed (NSU) Demand 6 / 20

Back to the UMP

▶ We can then reformulate the consumer’s UMP, incorporating

the budget constraint:

max u (x )
x ∈B(p,m)

▶ We have assumed preferences are continuous, and therefore

u (x ) is a continuous function.
▶ B (p, m) is a compact (closed and bounded) set.
▶ We must have x ≥ 0 and p.x ≤ m: the possible choices for x
are bounded.
▶ Because all of the constraints above are inequality constraints,
all of the boundaries are included in the set, hence the set is
closed.

Rafayal Ahmed (NSU) Demand 7 / 20

Extreme Value Theorem

▶ You should know the following important result, which I will

call the Weierstrass Maximum Theorem:

Theorem
(Weierstrass Maximum Theorem) Any continuous function on a
closed and bounded domain achieves its maximum and minimum.
Corollary
Suppose a consumer’s preferences are represented by the
continuous utility function u. Then her UMP has at least one
solution.

Rafayal Ahmed (NSU) Demand 8 / 20

Demand

▶ Having shown that the UMP has solution(s), let us now focus
on analyzing these solutions.
▶ “If you know something exists, give it a name”.
▶ For given preferences ≿, the solutions(s) to the UMP is a
mapping from (p, m) to X .
▶ Let us call this mapping the demand correspondence,
D : (p, m) ⇒ X .
▶ D maps (p, m) to a subset of X . This subset, denoted as
D (p, m), includes all solutions to the UMP.
▶ Because there may be more than one solution to the UMP, we
are calling D a correspondence.
▶ When the solution is unique, we could call D the demand
function.

Rafayal Ahmed (NSU) Demand 9 / 20

Demand with Convex Preferences

Theorem
Suppose the consumer’s preferences are convex. Then, the demand
correspondence D (p, m) is a convex set.
Proof.
Suppose x ∗ , y ∗ ∈ D (p, m). That is, both x ∗ and y ∗ are solutions to the UMP.
That means both of them must also be in the budget set, so x ∗ , y ∗ ∈ B (p, m).
Moreover, we must have u (x ∗ ) = u (y ∗ ) = maxx ∈B(p,m) u (x ), so x ∗ ∼ y ∗ and
thus x ∗ ≿ y ∗ . Now let α ∈ [0, 1] be arbitrary and consider the convex
combination αx ∗ + (1 − α) y ∗ . We have to show αx ∗ + (1 − α) y ∗ ∈ D (p, m).
Because B (p, m) is convex, we know αx ∗ + (1 − α) y ∗ ∈ B (p, m). Because
preferences are convex, αx ∗ + (1 − α) y ∗ ≿ y ∗ , so
u (αx ∗ + (1 − α) y ∗ ) ≥ u (y ∗ ) = maxx ∈B(p,m) u (x ). Thusly, αx ∗ + (1 − α) y ∗
is a solution to the UMP and αx ∗ + (1 − α) y ∗ ∈ D (p, m).

Rafayal Ahmed (NSU) Demand 10 / 20

Unique Solution to the UMP

Theorem
Suppose the consumer’s preferences are strictly convex. Then, the demand
correspondence D (p, m) is a singleton. That is, there is a unique solution to
the UMP and D (p, m) is a function.

Proof.
Towards a contradiction, assume there are two solutions x ∗ ̸= y ∗ to the UMP.
So x ∗ , y ∗ ∈ B (p, m) and u (x ∗ ) = u (y ∗ ) = maxx ∈B(p,m) u (x ), so x ∗ ≿ y ∗ .
Take α ∈ (0, 1). Because B (p, m) is a convex set,
αx ∗ + (1 − α) y ∗ ∈ B (p, m). Because preferences are strictly convex,
αx ∗ + (1 − α) y ∗ ≻ y ∗ , so u (αx ∗ + (1 − α) y ∗ ) > u (y ∗ ). Thus, we have
found a bundle in B (p, m) which has a strictly higher utility level than y ∗ . This
contradicts the fact that y ∗ is a solution to the UMP.

Rafayal Ahmed (NSU) Demand 11 / 20

Demand Function

▶ With strictly convex preferences, we cannot have two distinct

optimal bundles, because any convex combination of two
distinct bundles will be strictly preferred by the consumer.
Hence the unique solution to the UMP.
▶ If we assume strictly convex preferences, we will have a unique
solution to the UMP.
▶ It is reasonable to expect that any such solution will be on the
budget line, as opposed to the interior of the budget set.
▶ The budget line is the set {x ∈ X : p.x = m}.

Rafayal Ahmed (NSU) Demand 12 / 20

Demand with Monotonic Preferences
Proposition
Suppose ≿ are monotonic. Then for any x ∗ ∈ D (p, m), we must
have p.x ∗ = m.
Proof.
Suppose p.x ∗ ̸= m. Because we must have p.x ∗ ≤ m, here we will
have p.x ∗ < m. In that case, we can pick ϵ > 0 small enough such
that we have the bundle x ∗ + ϵ = (x1 + ϵ, ..., xL + ϵ) be such that
m−p.x ∗
p. (x ∗ + ϵ) ≤ m. In particular, any ϵ < P L would work.
i=1
pi
Because preferences are monotonic and x ∗ + ϵ >> x ∗ , we will have
u (x ∗ + ϵ) > u (x ∗ ), which contradicts the fact x ∗ ∈ D (p, m).

Remark
In words, this result means with monotonic preferences, the
consumer’s budget constraint must hold with equality, and thus
the optimal bundle must be on the budget line.
Rafayal Ahmed (NSU) Demand 13 / 20
D (p, m) is Homogenous of Degree Zero

Proof.
Suppose initially the price vector and income were (p, m). Let
λ > 0 be arbitrary. Consider what happens to the budget set when
all prices and income is multiplied by λ.
B (λp, λm) = {x : (λp) .x ≤ λm} = {x : p.x ≤ m} = B (p, m).
The consumer’s preferences are the same as before, and the
consumer has the same budget set to choose from, so clearly
D (λp, λm) = D (p, m). This, by definition, says that the demand
correspondence is homogenous of degree zero.

Rafayal Ahmed (NSU) Demand 14 / 20

Marginal Utility

▶ Now assume the consumer’s utility function is differentiable,

that is, the partial derivatives ∂u(x∂x
1 ,...,xL )
ℓ
exist for all
ℓ ∈ {1, ..., L}.

Definition
At bundle x , the consumer’s marginal utility for good i is:

∂u (x )
MUi =
∂xi

Proposition
Suppose the consumer’s preferences are (strongly) monotonic.
Then for all goods ℓ ∈ {1, ..., L}, the marginal utility MPℓ is
(strictly) positive.

Rafayal Ahmed (NSU) Demand 15 / 20

Marginal Rate of Substitution

▶ Suppose the consumer initially has bundle x = (x1 , ..., xL ).

▶ He gets additional amount ϵj > 0 of good j, and loses amount
ϵi > 0 of good i.
▶ Think of these as infinitesimal amounts.
▶ His new bundle is x̂ = (x1 , ..., xi − ϵi , ..., xj + ϵj , ..., xL ).
▶ We want to know the ratio ϵϵj such that x̂ ∼ x ; that is,
i
u (x̂ ) = u (x ).

Definition
ϵ
The (positive) ratio ϵji is called the marginal rate of substitution of
good i for good j, and written as MRSi,j .

Rafayal Ahmed (NSU) Demand 16 / 20

Marginal Rate of Substitution

▶ Instead of ϵ, let dxj and dxi be the changes in amounts of

good j and good i.
▶ These changes can be either positive or negative.
▶ Amount of every other good stays the same, so dxℓ = 0 for
ℓ ̸= i and ℓ ̸= j.
▶ We want to know the (positive) ratio dx j
dxi such that utility
level does not change due to these changes: du (x1 , ..., xL ) = 0
∂u
▶ Now, du (x1 , ..., xL ) = ∂x ∂u
j
dxj + ∂xi
dxi = 0. A simple bit of
algebra and we get:
∂u
dxj ∂xi MUi
MRSi,j = = ∂u
=
dxi ∂xj
MUj

Rafayal Ahmed (NSU) Demand 17 / 20

Back to the UMP

▶ Let us carefully state all of the constraints for the UMP, and
set it up as a constrained maximization problem (with
inequality constraints):

max u (x1 , ..., xL )

x1 ,...,xL
L
X
subject to: pℓ xℓ ≤ m ; (Budget Constraint)
ℓ=1
for all ℓ, xℓ ≥ 0 ; (Non-negativity)

Rafayal Ahmed (NSU) Demand 18 / 20

Kuhn-Tucker Conditions

▶ The Lagrangian for the UMP is:

L L
!
X X
L = u (x1 , ..., xL ) + λ m − pℓ xℓ + µℓ xℓ
ℓ=1 ℓ=1

▶ Here, λ is the multiplier for the budget constraint, and µℓ are

the multipliers for the non-negativity constraint.
▶ Recall that the complementary slackness conditions for the
non-negativity constraints mean for any ℓ ∈ {1, ..., L},
µℓ xℓ∗ = 0.
▶ This means whenever xℓ∗ > 0, we must have µℓ = 0.

Rafayal Ahmed (NSU) Demand 19 / 20

MRS = price ratio
▶ Suppose x ∗ ∈ D (p, m). Consider two goods i and j that are
consumed in the optimal bundle x ∗ .
▶ That is, xi∗ > 0 and xj∗ > 0. We must have µi = µj = 0.
▶ The FOCs for i and j give us:

∂u (x ∗ )
− λpi + µi = 0 (1)
∂xi
∂u (x ∗ )
− λpj + µj = 0 (2)
∂xj
▶ Dividing (1) by (2), we get:
∂u(x ∗ )
∂xi pi
∂u(x ∗ )
=
pj
∂xj

▶ This is the well-known MRS = price ratio result. Notice this

need not be true if either xi∗ or xj∗ is zero.
Rafayal Ahmed (NSU) Demand 20 / 20