ECN 145 Lecture 12

Transportation Economics:
Production and Costs I
Demand and Supply

• Chapters from Essays textbook:

P Supply (Chapter 3)

Pricing (Chapter 4)
P0
Demand (Chap. 2)

Q0 Q
How to obtain supply?

• 1)What is the production technology?

• 2)Therefore, what are production costs?
• 3)Given demand, what will the firm (or
government) choose to supply?
• We will focus here on production and
costs.
Production function:
• Given inputs x=(x1,x2,…,xn), write output y
as, y = f ( x ,..., x ) = f ( x )
1 n
• Properties:
– Increasing: ∂f / ∂x > 0
– Quasi-concave:
– if f ( x 0 ) = f ( x1 ) 0 ≤ λ ≤1
and
– then f (λ x 0 + (1 − λ ) x1 ) ≥ f ( x 0 )
Production Function

• Illustrate with “iso-quants”:

X2

x1

f(x)=y1>y0
λ x0+(1- f(x)=y0
λ )x1 x0
X1
Returns to Scale
• Suppose that there is one input x, and,
α
y = f (x) = x
∀ α =1 ⇒ constant returns to scale
• doubling all inputs will just double output
∀ α >1 ⇒ Increasing returns to scale
• doubling all inputs more than doubles output
∀ α <1 ⇒ decreasing returns to scale.
• doubling all inputs will less than double output
Input prices

• Prices for inputs xi are wi

• E.g. labor – wage
• Capital – rental price
• Fuel – cost of oil;
• Total costs are,
n
C = ∑w i x i
i =1
The firm’s problem:
n
min ∑ wixi subject to f(x) > y
i =1
x2

B Slope=-w1/w2

A
C f(x)=y
x1
The firm’s problem (cont’d):

• Point A is the lowest cost method of

producing y
– B and C are more expensive to get y
• Write solution as cost function:
n
C ( y, w ) = ∑ w i x i ( y, w )
• with: i =1
• x i ( y, w ) - input demands
Change in Price:

• Suppose price falls from w1 to w1’:

x2
Slope = -w1/w2

f(x)=y
B Slope = -w1′ /w2

x1
Change in Price (cont’d):

• Fall in w1 will increase demand for x1,

and reduce demand for x2, moving from
A to B.
∂x i
• < 0 - pure “substitution” effect
∂w i
Demand Curve
• This gives us downward sloping demand:

wi

D
Xi
Derivative of costs:

∂C n ∂x j
= x i ( y, w ) + ∑ w j = x i ( y, w )
∂w i j=1 ∂w i
• Because
n ∂x j
∑ w j ∂w =0
j=1 i
– (substitution effects all cancel out)

• Thus, the derivative of costs w.r.t. factor prices

equals factor demands
Returns to Scale:
α
• If, y = f (x) = x
1/ α
• then, C ( y, w ) = y w
∀ α =1 ⇒ doubling output will double costs;
∀ α >1 ⇒ Increasing returns to scale.
• doubling output will less than double costs;
∀ α <1 ⇒ Decreasing returns to scale.
• doubling output will more than double costs.
Average costs:

• Hold the (single) input price w fixed:

• Define average costs,

1/ α 1−α
C( y, w ) y w
AC = = = wy α ,
y y
• E.g. Total costs =$100, y=5, so AC=$20
Marginal costs

• Hold the (single) input price w fixed:

• Define marginal costs,

1−α
∂C( y, w ) w
MC = = y α
∂y α
• E.g. Total costs=$100 when y=5, $115 when y=6
• So marginal costs are $15.
Returns to Scale:

1−α 1−α
1
AC = wy α , MC = wy α
α
• so, AC = α = Total Costs ,
MC y ⋅ MC
• which is a measure of returns to scale!
Constant Returns to Scale:
$
α= Constant
returns
1
φ AC=MC

y
Increasing Returns to Scale:
$
α >1
Increasing
returns

AC

MC

y
Decreasing Returns to Scale:
$ Decreasing
MC
α <1 returns
(Say,
α =1/2)
AC

y
Eg: Cobb-Douglas Production Function
α β
y = f (x) = x1 x 2 , α, β > 0
• Notice that doubling both x1 and x2:

α β α +β
f (2 x ) = (2 x1 ) (2 x 2 ) = 2 f (x)

• So,
• α + β = 1⇒ constant returns to scale
• α + β > 1⇒ increasing returns to scale
• α + β < 1⇒ decreasing returns to scale
Cobb-Douglas Cost Function
α β
min w1 x1 + w 2 x 2 subject to x1 x 2 ≥y
• Use a Lagrangian,

L = w1 x1 + w 2 x 2 + λ (y - x1α x β2 )
• Find first-order condition w.r.t x1, x2:
∂L α −1 β
= w 1 − λ xα1 x 2 = 0 ⇒ w 1x 1 = λ α
f (x)
∂x1
∂L
= w 2 − λ βx1α x β2−1 = 0 ⇒ w 2 x 2 = λ βf ( x )
∂x 2
Cobb-Douglas Cost Function (cont’d)
• From these two conditions, we solve for,

w1 x1 + w 2 x 2 = λ(α + β)f ( x )
• We can solve for λ from the FOC,

β β β β β
w1α x1α α α α
= λ α f (x) , w 2x 2 = λ β f (x)
α β α +β α β α +β
⇒ ( w 1 w 2 )f ( x ) = λ (α β )f ( x )
Cobb-Douglas Cost Function (cont’d)
• Solving for λ , costs then are,

C( y, w1, w 2 ) = w1 x1 + w 2 x 2
1 α β
= λ (α + β )f ( x ) = (α + β )Ay α +β
( w1α +β w 2α +β )

• where,
α β
• − − is a constant.
α+β α+β
A = (α β )
Returns to Scale:
• First write the log of costs as:

 1   α   β 
ln C = B +   ln y +   ln w1 +   ln w 2
 α +β  α +β  α +β
• Differentiating this w.r.t. y, we see that,

−1
AC Costs  ∂ ln C 
= =   = (α + β)
MC y(∂C / ∂y)  ∂ ln y 
• measures the returns to scale!
Cobb-Douglas Input Demands:
• Demands for inputs are obtained by
differentiating costs:
1 α −1 β
∂C α +β α +β α +β
x1 ( y, w 1 , w 2 ) = = α Ay w1 w 2
∂ w1
1 α β −1
∂C α +β α +β α +β
x 2 ( y, w1, w 2 ) = = β Ay w1 w 2
∂w 2
Costs Shares:
• Comparing x1 and x2 with total costs:
1 α β
α +β α +β α +β
C = (α + β)Ay ( w1 w 2 )
• we see that,
w1x1 α
=
C α +β
• which are constant!
w 2x 2 β
=
C α+β
Elasticity of Substitution
• For the Cobb-Douglas function:
x1 α w 2
=
x 2 β w1
• So the elasticity of substitution is,

∂ ln( x1 / x 2 )
− =1
∂ ln( w1 / w 2 )

• This may not be a good description of actual

substitution between inputs! So consider…..
Translog Cost Function

• Many outputs (joint prod.) y =( y1 ,, y m )

• And inputs w = ( w1 ,  , w n )
m n
ln C( y, w ) = a 0 + ∑a i ln y i + ∑b i ln w i
i =1 i =1

1 m n 1 m n
+ ∑∑a ij ln y i ln y j + ∑∑b ij ln w i ln w j
2 i =1 j=1 2 i =1 j=1

m n
+ ∑∑g ij ln y i ln w j
i =1 j=1
Translog Cost Function (cont’d)
• Note that the first line is just Cobb-Douglas,
• (in logs, with multiple inputs and outputs):

m n
ln C( y, w ) = a 0 + ∑ a i ln yi + ∑ bi ln w i
i =1 i =1
• The “extra” terms on the second and third lines allow for
more general substitution between inputs and outputs.
Translog Cost Shares:
• Differentiating the cost function,

∂ ln C x i w i n m
= = bi + ∑ b ij ln w j + ∑ g ij ln y j
∂ ln w i C j=1 i =1
• so that, 2
∂ ln C
= bij ≠ 0
∂ ln w i ∂ ln w j
• this allows for a wide pattern of substitution
between inputs.
Returns to Scale:
−1
AC Costs  
= =  ∑ ∂ ln C / ∂ ln yi 
MC y ⋅ (∂C / ∂y)  i 
−1
• so if aij =0, then, AC  
=  ∑ a i 
MC  i 

• is a measure of returns to scale!