04.10.

202
Microeconomic Analysi
[email protected]
[email protected]

• Ulusal Toplu Katalo

• abebook.com
• Simon Blume - Mathematics for Economist

11.10.202
Microeconomic Analysi
For understanding dependent/independent variables, it’s useful to draw functions on coordinate
plane

Pro t
Pro t

Q Q

Linear/Non-Linear functions
We use continuous functions and skip non-continuous relationships. If relationship is continuous,
we use derivatives. This comes from economic principles because people always act on margin.
f (x) − f (c)
f ′(c) = li mx→c
x −c
All alternatives have limit, so we need to choose our
• f (x) = x r - Elementary power rule / Polynomial rule
f′(x) = r x r−1


fi
fi
.

 • h(x) = a f (x) + bg(x)

h′(x) = a f′(x) + bg′(x)
• (a f )′ = a f′ - Constant factor rule
• ( f ± g)′ = f′ ± g′ - Sum / Difference rule
• h(x) = f (x)g(x)
h′(x) = f′(x)g(x) + f (x)g′(x) - Product rule
• h(x) = f (g(x))
h′(x) = f′(g(x))g′(x) - Chain rule
• g( f (x)) = x
f (g(y)) = y
1
g′(x) =
f′(g(x))
dx 1
= dy - Inverse function rule
dy
dx
1
h =
• f (x)
f′(x)
h′(x) = − - Reciprocal rule
f (x)2

• (g)
′
f f′g − fg′
= - Quotient rule
g2
• Generalized power rul
d ax ax
• d x (c ) = c lnc ⋅ a
• y = ax

Green line represents e x = y, blue line represents 3x = y, and red line corresponds to 2x = y
y = ax ex = y
loga y = loga(a x ) loge y = loge(e x )
loga y = xloga a lny = x
loga y = x
By taking derivatives, we can scale data which is random walk
y = ax y = a + bx





















e

 a 1
loga y = x (1) x =− + y (2
b b
(1) and (2) are inverse functions

Cobb-Dougla
y = x1α x2β
lny = αln x1 + βln x2
per centageΔinQ
Price elasticity of demand: ϵ=
per centageΔin P
ΔQ /Q ΔQ P
ϵ= = ⋅
ΔP/P ΔP Q
slope rati
When we use ln scale, the coef cients will be the elasticities
dlny
ϵ= =α
dln x1
dlny
ϵ= =β
dln xx
d ax ax
• d x (e ) = a e
• y = ex
y′ = e x
d 1
• dx (log c x) = c>0, c≠
xlnc
d 1
• d x (ln x) = x
d x x
• d x (x ) = x (1 + ln x)

One-variable case

In order to be an optimum point, slope=0. In the rst graph, we have minimum, while in the second
one, we have maximum point
f (x) = 3x 2


fi
1

fi
.

 f′(x) = 6x = 0 - necessary
x* = 0
f′′(x) = 6 > 0 - The function is suf cient because 2nd order derivative is more than 0, which means
function should increase further. Thus, x* is minimum point
z = f (x, y) = − (x 2 + y 2 ) + 4
(x, y, z) = (0,0,4)

FOC
∂f (x, y)
• =0 fx = 0 fx = − 2x = 0 x* = 0
∂x
∂f (x, y)
=0 fy = 0 fy = − 2y = 0 y* = 0
• ∂y

f (x1, x2 )
∂f (x1, x2, . . , xi )
f′i = = ∇f (xi ) = 0
∂xi
SOCs (Linear Algebra

[ 21 a 22]
a11 a12
A= a

[ f21 f22]
f11 f12
H= - Hessian Matrix

We need to check whether | H | is positive de nite or negative de nite for nding the optimum
point of graph
∂ 2 f (x1, x2 )
f11 =
∂x12
∂ 2 f (x1, x2 )
f12 = = fxy
∂x1∂x2
Young's Theorem
According to Young’s Theorem, cross partial derivatives of functions are equal
f12 = f21
2
∂ f ∂2 f
=
∂x1∂x2 ∂x2∂x1
For Hessian Matrices
• If f11 > 0 and f11 f22 − f12 f21 ≥ 0 (>0), the matrix is positive semi-de nite (de nite) - minimum
poin
• If f11 < 0 and f11 f22 − f12 f21 ≥ 0 (>0), the matrix is negative semi-de nite (de nite) -
maximum poin

Let’s look at our example

[ 0 −2]
−2 0
H= , then f11 = − 2 < 0 and | H | = (−2)(−2) − 0 = 4 ≥ 0




t

fi
fi
.

fi
fi
.

fi
fi
fi
fi
 Thus, our example is negative semi-de nite, which means (x*, y*, z*) = (0,0,4) is the maximum
point of the function

Constrained Optimization

y ω= budget constrain
ω Everyone can have different utility, but we
Py need to nd optimum. Of course, more is
better, but our desires are constrained with
budget
y*

un

u(x, y)

x* ω
x
Px

m a xU(x, y) = x y
s.t
g(x, y) = con sta nt
ω = Px ⋅ x + Py ⋅ y
ω Py
x= − y (1
Px Px
So, by using (1) transformation, we can easily maximize our utility (used by Varian)

( Px Px )
ω Py
m a x yU(x, y) = − y y

Langrange Metho
ℒ = U(x, y) + λ(c − g(x, y))
ℒ = U(x, y) − λ(g(x, y) − c)
ℒ = x y + λ(ω − Px x − Py y)
FOC
ℒ(x, y, λ)
y x
ℒx = 0 ℒx = y − Px λ = 0 λ= =
• Px Py
Px
ℒy = 0 ℒy = x − Py λ = 0 y= x
• Py
Px x
ℒλ = 0 ℒλ = ω − Px x − Py y = 0 ω − Px x − Py =0
• Py
ω − 2Px x = 0
ω ω
x* = and y* =
2Px 2Py

fi
fi
t

 x* and y* are demand functions and they’re symmetric functions

SOCs
ℒλλ ℒλ1 ℒλ2 0 −Px −Py
| Hb | = ℒ1λ ℒ11 ℒ12 = −Px 0 1
ℒ2λ ℒ21 ℒ22 −Py 1 0

Red part is A, Blue part is A T, Green part is Hessian Matrix without constraints, and Orange part is
borders de ned with λ
det | Hb | ≥ 0 → negative semi-de nite, thus maximum point

ℒ(x1, . . . , xn, μ1, . . . , μk ), when we have n variables and k constraints

• n + k - FOC
• (n + k) × (n + k) - SOC
u(x, l )
given 2 constraint
Px x = (16 − l )h + w
l = 16
ℒ = u(x, l ) + λ( . . . ) + μ( . . . )
ℒλλ ℒλμ A
| Hb | = ℒμλ ℒμμ A
AT AT H
This pink part will be equal to 0s

18.10.202
Microeconomic Analysi
• Mathematic
• Sorting the goods/service
• Number Theor
P
• Rational Number Choice (r = q division of 2 integers
1. Completeness - comes from the consumer tests. Consumer knows everything according to
completeness axiom. (Full Information
x > y - Prefer x to
x ∼ y - I’m indifferent
x < y - Prefer y to
2. Transitivit
x > y, y > z, then x > z
Example: 3 second-hand cars in the marke
FOCU
ACCEN
JETT
If there is someone who owns ACCENT
A

fi
1

fi
)

 A>F, F>J, but J>A → intransitive, inconsistent, irrational

100 100 10
Sometimes we change our preferences because of environment y

There are different personalities that are still rational x2

• Self-regarding, altruisti f (x)

Ω+ x

• Self-interested x1

Ω-
Preference Utilit
We have several assumptions for having a utility function. In this theory, we take into the account
the binary relationships: “⪰” → x1 ⪰ x 2 → “x1 is at least as good as x 2”
Axiom 1: Completeness - For all x1 and x 2 in X , either x1 ⪰ x 2 or x 2 ⪰ x1 . Person has ability to
discriminate and the necessary knowledge to evaluate alternatives
Axiom 2: Transitivity - Consistency - For any three elements x1 , x 2 , and x3 in X , if x1 ⪰ x 2 and
x2 ⪰ x3, then x1 ⪰ x3. Some economists have proven that
These two axioms together imply that the consumer can completely rank any nite number of
elements in the consumption set, X, from best to worst, possibly with some ties.

De nition 1: The binary relation ⪰ on the consumption set X is called a preference relation if it
satis es Axioms 1 and 2.

De nition 2: The binary relation ≻ on the consumption set X is de ned as follows:

x1 ≻ x2 if and only if x1 ⪰ x2 and x2 ⪰̸ x1.

The relation ≻ is called the strict preference relation induced by ⪰ , or simply the strict preference
relation when ⪰ is clear. The phrase x1 ≻ x 2 is read, ‘x1 is strictly preferred tox 2’.

De nition 3: The binary relation ∼ on the consumption set X is de ned as follows:

x1 ∼ x2 if and only if x1 ⪰ x2 and x2 ⪰ x1

The relation ∼ is called the indifference relation induced by ⪰ , or simply the indifference relation
when ⪰ is clear. The phrase x1 ∼ x 2 is read, ‘x1 is indifferent to x 2’

De nition 4: Let x0 be any point in the consumption set, X. Relative to any such point, we can
de ne the following subsets of X:

1. ⪰ (x0 ) ≡ (x | x ∈ X, x ⪰ x0 ), called the ‘at least as good as’ set.

2. ⪯ (x0 ) ≡ (x | x ∈ X, x ⪯ x0 ), called the ‘no better than’ set.

3. ≺ (x0 ) ≡ (x | x ∈ X, x ≺ x0 ), called the ‘worse than’ set.

fi
fi
fi
fi
fi
fi

fi
fi
.

fi

 4. ≻ (x0 ) ≡ (x | x ∈ X, x ≻ x0 ), called the ‘preferred to’ set

5. ∼ (x0 ) ≡ (x | x ∈ X, x ∼ x0 ), called the ‘indifference’ set.

Axiom 3: Continuity - For all x ∈ R+n , the ‘at least as good as’ set, ⪰ (x), and the ‘no better than’
set, ⪯ (x), are closed in R+n .

Axiom 4: Local Non-satiation - For all x0 ∈ R+n , and for all ϵ > 0 , there exists some
x ∈ Bϵ(x0 ) ∩ R+n , such that x ≻ x0.

Axiom 5: Monotonicity - For all x0 , x1 ∈ R+n , if x0 ≥ x1 then x0 ⪰ x1 , while if x0 ≫ x1 , then

x0 ≻ x1.

Axiom 6: Convexity - If x1 ⪰ x0, then t x1 + (1 − t)x0 ⪰ x0 for all t ∈ [0,1].

After satisfying all these axioms, we can have utility function

Indifference Curve Map

Marginal rate of substitution of good two for good one is the slope of indifference curve and shows
the preferences of people. Indifference curves should follow the principle of diminishing MRS.

 Utility function
De nition 5: A real-valued function u : R+n → R is called a utility function representing the
preference relation ⪰ , if for all x0, x1 ∈ R+n , u(x0 ) ≥ u(x1) ⟺ x0 ⪰ x1.

Consumer’s Problems
• Cobb-Dougla
• CES-type utility function (Constant Elasticity Function)
Budget: ω = p1 x1 + p2 x 2
ω p2 x2
x1 = −
p1 p1
since the functions are symmetrical:
ω p1 x1
x2 = −
p2 p2
m a xU(x1, x2 )
s.t. Consumer’s Optimization Proble
ω = p1 x1 + p2 x2

x*
1
is a function p1 which means when p1 ↑ , x1 ↓ . This satis es Law of Demand and this function
of x*
1
is called Marshallian Demand Function
ℒ = u(x1, x2 ) + λ(ω − p1 x1 − p2 x2 )

∂ℒ ∂u
= − λp1 = 0
∂x1 ∂x1

∂ℒ ∂u
= − λp2 = 0
∂x2 ∂x2

∂ℒ
= ω − p1 x1 − p2 x2 = 0
∂λ
∂u /∂x1 ∂u /∂x2
λ= =
p1 p2

MU1 p
MRSx1,x 2 = = 1
MU2 p2

MU1 MU2
λ= =
p1 p2

u = x1α x21−α

s.t. ω = p1 x1 + p2 x 2

ℒ = x1α x21−α + λ(ω − p1 x1 − p2 x2 )

fi
s

fi

 ∂ℒ
= α x1α−1 x21−α − λp1 = 0
∂x1

∂ℒ
= (1 − α)x1α x2−α − λp2 = 0
∂x2

α x1α−1 x21−α (1 − α)x1α x2−α

λ= =
p1 p2

α x1α−1 x21−α p1
=
(1 − α)x1α x2−α p2

α x2 p
= 1
(1 − α)x1 p2

p1(1 − α)x1
x2 =
p2 α

Use in 3rd derivative

p1(1 − α)x1 αx p (1 − α)p1 x1

ω − p1 x1 − p2 =ω− 1 1 −
p2 α α α
αω
x*
1
=
p1

∂x1 α
= > 0 - normal goo
∂ω p1

∂x1 αω
=− 2 <0
∂p1 p1

25.10.202
Microeconomic Analysi
m a xU(x)
s.t
ω = px
ℒ = U(x) + λ(ω − px)
FOC
∂ℒ
=0
∂x1
.
.
.

 ∂ℒ
=0
∂xn
∂ℒ
=0 Roy’s identit
∂λ
We nd optimum value of x*
1
, x*
2
, . . . , x*
n which are Marshallian Demand

u(x*( p, w)) is an indirect utility function

Theorem 1.6: Properties of the Indirect Utility Function

If u(x) is continuous and strictly increasing on R+n , then v( p, y) de ned in (1.12) is

n
1. Continuous on R++ × R+

2. Homogeneous of degree zero in ( p, y),

3. Strictly increasing in y

4. Decreasing in p

5. Quasiconvex in ( p, y).

Moreover, it satis es
6. Roy’s identity: If v( p, y) is differentiable at ( p 0, y 0 ) and ∂v( p 0, y 0 )/∂y ≠ 0, then

∂v( p 0, y 0 )/∂pi
xi( p 0, y 0 ) = − , i = 1,...,n .
∂v( p 0, y 0 )/∂y

m a x f (x) ming(x)

s.t. c = g(x) s.t. z̄ = f (x)

maximisation problem minimisation problem - Duality principle

The expenditure function
Theorem 1.7: Properties of the Expenditure Function
If u( . ) is continuous and strictly increasing, then e( p, u) de ned in (1.14) is
fi
,

fi



fi
fi
.

 1. Zero when u takes on the lowest level of utility in U,

n
2. Continuous on its domain R++ ×

3. For all p > > 0, strictly increasing and unbounded above in u,

4. Increasing in p

5. Homogeneous of degree 1 in p

6. Concave in p.

If, in addition, u( . ) is strictly quasiconcave, we have

7. Shephard’ s lemma: e( p, u) is differentiable in p at ( p 0, u 0 ) with p 0 > > 0, and

∂e( p 0, u 0 )
= xih( p 0, u 0 ), i = 1,2,...,n
∂pi

e( p, u) = px
s.t. u(x) ≥ ū
ℒ = px + μ(ū − u(x))
We will nd, x1h, x2h, . . . , xnh, μ, where x1h, x2h, . . . , xnh are Hicksian Demand Functions
x h( p, u) - utility affects Hicksian Deman
x( p, ω) - budget affects Marshallian Deman
Shephard’s Lemm
Maximisation problem → Marshallian Deman
m a xU(x) = x1 x2
s.t
ω = p1 x1 + p2 x2
ω ω
x*
1
= and x*2
=
2p1 2p2
ω2
u = v( p, ω) - indirect utility function →
4p1 p2
Minimisation problem → Hicksian Deman
min( p1 x1 + p2 x2 )
s.t
x1 x2 = u
ℒ = p1 x1 + p2 x2 + μ(u − x1 x2 )
ℒ1 = p1 − μ x2 = 0
ℒ2 = p2 − μ x1 = 0
ℒλ = u − x1 x2 = 0
p p
μ= 1 = 2
x2 x1
p2 x2
x1 =
p1
𝒰
.

fi

 p2 x2
u− x =0
p1 2
p1u
x2h =
p2
p2u
x1h =
p1
p2u p1u
e( p, u) = p1 x1h + p2 x2h = p1 + p2
p1 p2
e*( p, u) = 2 p1 p2u
∂e( p, u) 1 p2u p2u
Shephard’s Lemma: = x1h( p, u) = 2 × =
∂p1 2 p1 p1
1) e( p, v( p, y)) ≤ ω
ω2
2 p1 p2 =ω
4p1 p2
2) v( p, e( p, u)) = u
2( p1 p2u)2
=u
4p1 p2
Theorem 1.9: Duality Between Marshallian and Hicksian Demand Functions
Under Assumption 1.2 we have the following relations between the Hicksian and Marshallian
demand functions for p > > 0, y ≥ 0, u ∈ U, and i = 1,...,n:

1. xi( p, y) = xih( p, v( p, y)).

2. xih( p, u) = xi( p, e( p, u)).

ω 2 p1 p2u p2u
x1 = x1h = =
2p1 2p1 p1

Relationships between the functions

m a xU = x1 x2

s.t
ω = p1 x1 + p2 x2
ω ω ω2
x1 = x2 = v( p, ω) =
2p1 2p2 4p1 p2
min( p1 x1 + p2 x2 )
s.t
x1 x2 = ū
p2u p1u
x1h = x2h = e( p, u) = 2 p1 p2u
p1 p2
.

 ω̄2
When p1 ↑ → ω ↑ → ū, we would have anchor: ua =
4p¯1 p¯2
ω̄ 2
p¯2 u p¯2 4p¯ ω̄ p¯1
1 p¯2
x1h = = =
p1 p1 2 p1

ω̄ 2
p1u p1 4p¯ ω p1
1 p¯2
x2h = = =
p¯2 p¯2 2p¯2 p¯1
p1
p¯1 x1h + p¯2 x2h = ω 2 = ω̄
p¯1
S* = ω* − ω̄
9
Example: p¯1 = 4,ω = 100,p1 = 9, then: w * = 100 = 150. 150 is the new optimum budget
4
Example: p1 = 4,p2 = 5,w = 100, then: u = x1 x 2 → x*
1
= 12.5,x*
2
= 10 → u = 125
p2u 25 p1u
x1h = = x2h = = 5 p1
p1 p1 p2
25
ω = p1 × + p2 × 5 p1 = 50 p1
p1
When p1 = 9, we would have ω* = 150.
Substitution and Income Effect

There are 2 effects: p1 ↑ ↓ (price effect) and ω ↑ ↓ (income effect)

u = x1 x2
p1 = 4 p2 = 1 ω = 72 x*
1
=9 x*
2
= 36
p′1 = 9 p2 = 1 ω = 72 x1′* = 4 x2′* = 36



s

x1 x2 = 36 × 4 = 144 x2 = 4x1 → tangency conditio

4x12 = 144
x1 = 6 and x2 = 24
ω ω
Example: Find the expenditure function and the Hicksian Demand: v( p, ω) = +
p1 p2
v( p, e( p, u)) = u
e( p, u) e( p, u)
+ =u
p1 p2
ppu
e( p, u) = 1 2
p1 + p2
∂e( p, u) h
p22u
= x1 ( p, u) =
∂p1 ( p1 + p2 )2
p12u
x2h( p, u) =
( p1 + p2 )2
x1h( p, v( p, ω)) = x1( p, ω)
∂v /∂ω
Roy’s Identity: x1( p, ω) = −
∂v /∂p1
Example: u = ln(x1) + x 2
MU1 p
= 1
MU2 p2
1 p p
= 1 x1 = 2
x1 p2 p1

( p1 )
p
u = ln 2 + x2

( p1 )
p
x2h = u − ln 2

x2h = u − ln( p2 ) + ln( p1) → goods are substitutes

Example: u = ln x1 + 3ln x 2 + ln(x3 + 1)
1
(3)
P= 2

Budget constraint: x1 + 2x 2 + 3x3 = ω

ℒ = ln x1 + 3ln x2 + ln(x3 + 1) + λ(ω − x1 − 2x2 − 3x3)
1 1
ℒ1 = −λ =0 x1 =
x1 λ
3 3
ℒ2 = − 2λ = 0 x2 =
x2 2λ
1 1
ℒ3 = − 3λ = 0 x2 = −1
x3 + 1 3λ
ℒ3 = ω − x1 − 2x2 − 3x3 = 0

( 3λ )
1 3 1
ω= + ×2+ −1 ×3
λ 2λ
.

 5
−3=ω
λ
5
λ=
ω +3
ω +3
x1 =
5
Theorem 1.11: The Slutsky Equation
Let x( p, y) be the consumer’ s Marshallian demand system. Let u* be the level of utility
the consumer achieves at prices p and income y. Then,

01.11.202
Microeconomic Analysi
Inputs → Outputs → This transformation is described in production function
Total Re venu e = P × Q
The most general way is to think of the rm as having a production possibility set,
Y ⊂ R m, where each vector y = (y1, . . . , ym) ∈ Y is a production plan whose
components indicate the amounts of the various inputs and outputs.

m a xπ = TR − TC
It is more convenient to describe the rm’s technology in terms of a production function.

PPF (Production Possibility Frontier) PPS (Production Possibility Set)

Assumption 3.1: The production function, f : R+n → R n , is continuous, strictly increasing, and
strictly quasiconcave on R+n , and f (0) = 0.
For any xed level of output, y, the set of input vectors producing y units of y
f (x)
output is called the y-level isoquant. -> like indifference curv

In general, there more than 2 inputs used in production which gives us multi-
dimensional graphics. But they are hard to deal with, so we decrease the
amount of these inputs until minimum which would be more or less realistic
x1 x2
fi
1

fi

fi
e

 y
x2
We x y level from 3D and turn it into 2

x1

y = ā
x2
x1

Isoquant - Isoquant Mappin

Euler Theorem: f (x) is homogeneous of degree k if and only if

n
∂f (x)
∑ ∂xi i
k f (x) = x for all x.
i=1

MRTS or MRT (Marginal Rate of Technical Transformation

∂f (x)/∂xi
MRTSij (x) =
∂f (x)/∂xj

MRTS is the slope of isoquant

De nition 3.1: Separable Production Functions

Let N = (1,...,n) index the set of all inputs, and suppose that these inputs can be partitioned into
S > 1 mutually exclusive and exhaustive subsets, N1, . . . , Ns . The production function is called
weakly separable if the MRTS between two inputs within the same group is independent of inputs
used in other groups:
∂( fi(x)/fj (x))
=0
∂xk for all i, j ∈ Ns and k ∉ Ns,

where fi and fj are the marginal products of inputs i and j . When S > 2 , the production function is
called strongly separable if the MRTS between two inputs from any two groups, including from the
same group, is independent of all inputs outside those two groups:
∂( fi(x)/fj (x))
=0
∂xk for all i ∈ Ns, j ∈ Nt and k ∉ NS ∪ Nt

De nition 3.2: The Elasticity of Substitution

For a production function f (x) , the elasticity of substitution of input j for input i at the point
x 0 ∈ R++
n
is de ned as
fi
fi
fi
fi

 ( )
dln MRTSij (x(r)) −1
0
σij (x ) ≡
dlnr
r=xj0 /xi0

where x(r) is the unique vector of inputsx = (x1, . . . , xn ) such that (i )xj /xi = r, (ii ) = xk = xk0 for
k ≠ i, j, and (iii )f (x) = f (x 0 ).
ΔQ /Q ΔQ P
ϵ= = → Price elasticity of demand is consisted of slope and ratio
ΔP/P ΔP Q
dlnQ
ϵ=
dln P
Log-Log relationship - y = α + βx + ϵ - Here, β gives the elasticity of x to y

Example: Q = a P −b
dQ
= − a bP −b−1
dP
ΔQ P P
ϵ= = − a bP −b−1 × =−b
ΔP Q a P −b
lnQ = ln a − bln P
dlnQ
=−b
dln P
ϵ → ∞ - P effect Substitutes 0 < ϵ < ∞ - Cobb-Douglas ϵ → 0 - Perfect Complement

CES type of production: ρ → 0,σij → 1 - Cobb-Douglas Production Function - y = Πxiα

ρ → − ∞, σij → 0 - Leontieff Production Function - y = min(x1, . . . , xi )

ρ → 1,σij → ∞ - Perfect Substitutes Production Function y = α x1 + . . . + γ xi

Theorem 3.1: (Shephard) Homogeneous Production Functions are Concave

Let f (x) be a production function satisfying Assumption 3.1 and suppose it is homogeneous o

 degree α ∈ (0,1]. Then f (x) is a concave function of x.

Returns To Scal
We need to classify time periods

1. Short-run - it is accepted as process. In short-run, we can’t change some inputs. Each economy
has its own short-run duration

2. Long-tun - long-run is generally accepted as a moment. The moment happens when a rm

enters new short-run
∂f (x)
MPi(x) = - Marginal Productivity
∂x
f (x)
A Pi = - Average Productivity
x
MPi Δ f (x)/f (x) Δ f (x) x
ϵi = = = × - Output Elasticity of Output
A Pi Δx /x Δx f (x)

MPi A Pi

De nition 3.3: (Global) Returns to Scale

A production function f (x) has the property of (globally):

1. Constant returns to scale if f (t x) = t f (x)for all t > 0 and all x;

2. Increasing returns to scale if f (t x) > t f (x) (x) for all t > 1 and all x;

3. Decreasing returns to scale if f (t x) < t f (x) for all t > 1 and all x.

De nition 3.4: (Local) Returns to Scale - Elasticity of Scal

The elasticity of scale at the point x is de ned as

n
dln[ f (t x)] ∑ fi(x)xi
μ(x) = limt→1 = i=1 ,
dln(t) f (x)

Returns to scale are locally constant, increasing, or decreasing as μ(x) is equal to, greater than, or
less than one. The elasticity of scale and the output elasticities of the inputs are related as follows:
n

∑
μ(x) = μi(x)
i=1

Example: y = x1α x2β

f (t x) = (t x1)α(t x2 ) β = t α+β x1α x2β

α + β = 1 → CRS - y = x1 x2 = x11/2 x21/2

α + β > 1 → IRS - y = x1 x2
fi
fi
.

fi



e

fi

 α + β < 1 → DRS - y = 3 x1 x2 = x11/3 x21/3

u(x, y) = f (x) + y - quasilinear function with corner solution

08.11.202
Microeconomic Analysis
c(w, y) - w is input prices, y is output
c(w, x) - w is input prices, x is inputs - Direct Cost Functio
c(w, x) = wx = w1 x1 + . . . + wn xn
De nition 3.5: The Cost Function
The cost function, de ned for all input prices w > > 0 and all output levels y ∈ f (R+n ) is the
minimum-value function

c(w, y) ≡ minx∈R+n (wx) s.t. f (x) ≥ y

If x(w, y) solves the cost-minimisation problem, then

c(w, y) = w × x(w, y)

min(wx)
s.t
ȳ = f (x)

AC,

MC AC
x1
MC

AFC b c

Y x2

∂ℒ
ℒ = wx + μ( ȳ − f (x)) μ= = MC xi = xi(w, y) - Conditional Input Deman
∂y
∂f (x)
ℒi = wi − μ* =0
∂xi
∂ℒ
ℒ = u(x) + λ(ω − px) λ= xi = xi( p, ω)
∂ω
Example: min(w1 x1 + w2 x 2 )
s.t. y = x1α x21−α
ℒ = w1 x1 + w2 x2 + μ(y − x1α x21−α )
∂ℒ w1 α x1α−1 x21−α
= w1 − μα x1α−1 x21−α = 0 μ= =
∂x1 w2 (1 − α)x1α x2−α
.

fi
1

fi
.

∂ℒ w1 α x2
= w2 − μ(1 − α)x1α x2−α = 0 μ= =
∂x2 w2 (1 − α)x1
∂ℒ α w2 x2
= y − x1α x21−α = 0 x1 =
∂μ (1 − α)w1

( (1 − α)w1 ) 2
α
α w2 x2
y− x 1−α = 0

( (1 − α)w1 ) 2
α
α w2 x2
x2(w, y) → y = x 1−α →

x1(w, y) →
Theorem 3.2: Properties of the Cost Function
If f is continuous and strictly increasing, then c(w, y) is

1. Zero when y = 0

2. Continuous on its domain

3. For all w > > 0, strictly increasing and unbounded above in y,

4. Increasing in w

5. Homogeneous of degree one in w

6. Concave in w.

Moreover, if f is strictly quasiconcave we have

7. Shephard’s lemma: c(w, y) is differentiable in w at (w 0, y 0 ) whenever w 0 > > 0, and

∂c(w 0, y 0 )
= xi(w 0, y 0 ), i = 1,...,n
∂wi

∂C ∂ 2C ∂ 2C
= xi → = → According to Young Theorem.
∂wi ∂wi∂wj ∂wj ∂wi

Theorem 3.3: Properties of Conditional Input Demands

Suppose the production function satis es Assumption 3.1 and that the associated cost function is
twice continuously differentiable. Then

1. x(w, y) is homogeneous of degree zero in w,

2. The substitution matrix, de ned and denoted

∂x1(w, y) ∂x1(w, y)
∂w1
... ∂wn
. . .
σ * (w, y) = . . .
. . .
∂x n (w, y) ∂x n (w, y)
∂w1
... ∂wn

is symmetric and negative semide nite. In particular, the negative semide niteness property implies

fi
,

fi
fi

fi

 ∂xi(w, y)
that ≤ 0 for all i.
∂wi

De nition 3.6: The Short-Run, or Restricted, Cost Function

Let the production function be f (z), wherez ≡ (x, x̄). Suppose that x is a subvector of variable
inputs and x̄ is a subvector of xed inputs. Let w and w̄ be the associated input prices for the
variable and xed inputs, respectively. The short-run, or restricted, total cost function is de ned as

sc(w, w̄, y; x̄) = minx (wx + w̄ x̄) s.t. f (x, x̄) > y

If x(w, w̄, y; x̄) solves this minimisation problem, then

sc(w, w̄, y; x̄) = wx(w, w̄, y; x̄) + w̄ x̄

The optimised cost of the variable inputs, w × x(w, w̄, y; x̄), is called total variable cost. The cost
of the xed inputs, w̄ x̄, is called total xed cost.

Short-run - process → some inputs are xe

Long-run - momen

Example: Pegasus has 200 airplanes - 1/1/202

It plans to produce 100 more airplanes in 2 years later - 300 airplanes - 1/1/202

2 years is a process and referred as short-run. While 1/1/2023 is the moment of long-run. After
1/1/2023, Pegasus will start new process/enter new short-run with K = 300

Example: ℒ = w1 x1 + w2 x 2 + μ(y − x1 x 2 )
ℒ1 = w1 − μ x2 = 0
AC
ℒ2 = w2 − μ x1 = 0
ℒμ = y − x1 x2 = 0
w x
μ= 1 = 2
w2 x1
w1 x1
x2 =
w2 AC
wx
y − x1 × 1 1 = 0
w2 Y
fi
fi
fi
t

fi
fi
.

fi
d

fi

 y w2 y w1
x1 = and x2 =
w1 w2
y w2 y w1
c(w, y) = w1 + w2
w1 w2
c(w, y) = 2 y w1w2 - Total Cos
2 y w1w2 2
AC = = - Assume w1 = w2 = 1.
y y
As we can see, there is increasing returns to scale (α = 1,β = 1,α + β > 1) since AC always
decreases
y = x1 x2
t y = (t x1)(t x2 ) = t 2 X1 x2
Example: ℒ = w1 x1 + w2 x 2 + μ(y − x1 x2 )
ℒ1 = w1 − 0.5μ x1−0.5 x20.5 = 0
ℒ2 = w2 − 0.5μ x10.5 x2−0.5 = 0
ℒμ = y − x1 x2 = 0
w1 w2
μ= =
0.5x1−0.5 x20.5 0.5x10.5 x2−0.5
w1 x w x
= 2 x1 = 2 2
w2 x1 w1
w2 x2
y − x2 × =0
w1
w2
y − x2 =0
w1

( w1 ) ( w2 )
0.5 0.5
w2 w1
x2 = y and x2 = y

( w2 ) ( w1 )
0.5 0.5
w1 w2
c(w, y) = w1 y + w2 y = 2y w1w2

Let’s assume w1 = w2 = 1
c(w, y) = 2y
AC = MC = 2
Since α + β = 1, we have constant returns to scale
% ΔinTC ∂TC Q MC
ϵ= = = - Output Elasticit
% ΔinQ ∂Q TC AC
If ϵ = 1, we have CRS
If ϵ < 1, we have IRS
If ϵ > 1, we have DRS
% ΔinTP ∂TP xi MPi
ϵ= = =
% Δin xi ∂xi TP A Pi
.