Chapter 4.

2
COST MINIMIZATION &
COST CURVES
Dr. Nguyen Bich Diep
[email protected]
THE COST-MINIMIZATION PROBLEM

• Consider a firm using two inputs to make one output.

• The production function is y = f(x1,x2).
• Take the output level y  0 as given.
• Given the input prices w1 and w2, the cost of an input bundle (x1,x2) is
w1x1 + w2x2.
THE COST-MINIMIZATION PROBLEM

• For given w1, w2 and y, the firm’s cost-minimization problem is to solve

min w 1x1 + w 2x 2
x1 , x 2  0

subject to f ( x1 , x 2 ) = y.
THE COST-MINIMIZATION PROBLEM

• The levels x1*(w1,w2,y) and x1*(w1,w2,y) in the least-costly input bundle

are the firm’s conditional demands for inputs 1 and 2.
• The firm’s cost function describes the minimum costs necessary to
achieve the desired level of output:
*
c( w 1 , w 2 , y ) = w 1x1 ( w 1 , w 2 , y )
*
+ w 2x 2 ( w 1 , w 2 , y ).
CONDITIONAL INPUT DEMANDS

• Given w1, w2 and y, how is the least costly input bundle located?
• And how is the total cost function computed?
ISO-COST LINES

• A curve that contains all of the input bundles that cost the same
amount is an iso-cost curve.
• E.g., given w1 and w2, the $100 iso-cost line has the equation

w 1x 1 + w 2 x 2 = 100 .
ISO-COST LINES

• Generally, given w1 and w2, the equation of the $c iso-cost line is

w 1x 1 + w 2 x 2 = c
i.e. w1 c
x2 = − x1 + .
w2 w2
Slope is - w1/w2.
ISO-COST LINES
x2 Slopes = -w1/w2.

c”  w1x1+w2x2

c’  w1x1+w2x2
c’ < c”

x1
THE Y’-OUTPUT UNIT ISOQUANT

• All input bundles yielding y’ units of output. Which is the cheapest?

x2

f(x1,x2)  y’

x1
THE COST-MINIMIZATION PROBLEM

• All input bundles yielding y’ units of output. Which is the cheapest?

x2

x2*

f(x1,x2)  y’

x1* x1
THE COST-MINIMIZATION PROBLEM

At an interior cost-minimization input bundle:

x2
* *
(a) f ( x 1 , x 2 ) = y  and
(b) slope of isocost = slope of isoquant
w1 MP1
− = TRS = − at ( x*1 , x*2 ).
w2 MP2
x2*

f(x1,x2)  y’

x1* x1
A COBB-DOUGLAS EXAMPLE OF COST
MINIMIZATION
• A firm’s Cobb-Douglas production function is
1/ 3 2 / 3
y = f ( x1 , x 2 ) = x1 x 2 .
• Input prices are w1 and w2.
• What are the firm’s conditional input demand functions?
A COBB-DOUGLAS EXAMPLE OF COST
MINIMIZATION
• At the input bundle (x1*,x2*) which minimizes
the cost of producing y output units:
(a) y = ( x* ) 1 / 3 ( x* ) 2 / 3
1 2
and

* −2 / 3 * 2 / 3
(b) w1  y /  x1 ( 1 / 3)( x1 ) (x2 )
− =− =−
w2  y /  x2 ( 2 / 3)( x*1 )1/ 3 ( x*2 ) −1/ 3
*
x2
=− .
*
2x1
A COBB-DOUGLAS EXAMPLE OF COST
MINIMIZATION *
w1 x 2
* 1/ 3 * 2 / 3 (b)
= .
(a) y = ( x1 ) ( x 2 ) w 2 2x*1
* 2w 1 *
x =
From (b) 2
x1 .
w2 2/ 3 2/ 3
* 1/ 3  2 w 1 *   2w 1 
Now substitute into (a) to get
y = ( x1 )  x1  =  x*1 .
 w2   w2 
2/ 3
*  w2 
So x1 =   y is the firm’s conditional demand for input 1.
 2w 1 
A COBB-DOUGLAS EXAMPLE OF COST
MINIMIZATION
2/ 3
2w 1 * *  w2 
*
Since x 2 = x1 and x1 =   y
w2  2w 1 

2/ 3 1/ 3
* 2w 1  w 2   2w 1 
x2 =   y=  y
w 2  2w 1   w2 
is the firm’s conditional demand for input 2.
A COBB-DOUGLAS EXAMPLE OF COST
MINIMIZATION
So the cheapest input bundle yielding y output units is

( x*1 ( w 1 , w 2 , y ), x*2 ( w 1 , w 2 , y ) )
  w  2/ 3  2w  1/ 3 
=  2  y,  1
 y .
  2w 1   w  
 2 
A COBB-DOUGLAS EXAMPLE OF COST
MINIMIZATION
So the firm’s total cost function is
c( w 1 , w 2 , y ) = w 1x*1 ( w 1 , w 2 , y ) + w 2x*2 ( w 1 , w 2 , y )
2/ 3 1/ 3
 w2   2w 1 
= w1   y + w2  y
 2w 1   w2 
2/ 3
 1
=  w1/ 3 2/ 3
w y + 21/ 3 1/ 3 2/ 3
w1 w 2 y
 2 1 2

1/ 3
 w 1w 2 
2
= 3  y.
 4 
A PERFECT COMPLEMENTS EXAMPLE OF
COST MINIMIZATION
• The firm’s production function is
y = min{ 4 x 1 , x 2 }.
• Input prices w1 and w2 are given.
• What are the firm’s conditional demands for inputs 1 and 2?
• What is the firm’s total cost function?
A PERFECT COMPLEMENTS EXAMPLE OF
COST MINIMIZATION
x2 Where is the least costly input bundle
4x1 = x2
yielding y’ output units?

x2* = y min{4x1,x2}  y’

x1* x1
= y/4
A PERFECT COMPLEMENTS EXAMPLE OF
COST MINIMIZATION
The firm’s production function is y = min{ 4 x 1 , x 2 }
and the conditional input demands are
*
*
x1 ( w 1 , w 2 , y ) =
y and x 2 ( w 1 , w 2 , y ) = y.
4
*
So the firm’s total cost function is c( w 1 , w 2 , y ) = w 1x1 ( w 1 , w 2 , y )
+ w 2x*2 ( w 1 , w 2 , y )
y  w1 
= w1 + w 2y =  + w 2  y.
4  4 
AVERAGE TOTAL PRODUCTION COSTS

For positive output levels y, a firm’s average total cost of producing y

units is

c( w 1 , w 2 , y )
AC( w 1 , w 2 , y ) = .
y
RETURNS-TO-SCALE AND ATC

• The returns-to-scale properties of a firm’s technology determine how

average production costs change with output level.
• Our firm is presently producing y’ output units.
• How does the firm’s average production cost change if it instead
produces 2y’ units of output?
CONSTANT RETURNS-TO-SCALE AND ATC

• If a firm’s technology exhibits constant returns-to-scale then doubling

its output level from y’ to 2y’ requires doubling all input levels.
• Total production cost doubles.
• Average production cost does not change.
DECREASING RETURNS-TO-SCALE AND ATC

• If a firm’s technology exhibits decreasing returns-to-scale then

doubling its output level from y’ to 2y’ requires more than doubling all
input levels.
• Total production cost more than doubles.
• Average production cost increases.
INCREASING RETURNS-TO-SCALE AND ATC

• If a firm’s technology exhibits increasing returns-to-scale then

doubling its output level from y’ to 2y’ requires less than doubling all
input levels.
• Total production cost less than doubles.
• Average production cost decreases.
RETURNS-TO-SCALE AND ATC
$/output unit

AC(y)
decreasing r.t.s.

constant r.t.s.

increasing r.t.s.

y
RETURNS-TO-SCALE AND TOTAL COSTS

• What does this imply for the shapes of total cost functions?
RETURNS-TO-SCALE
$
AND TOTAL COSTS
c(y)

c(2y’) Slope = c(2y’)/2y’

= AC(2y’).

Slope = c(y’)/y’
= AC(y’).

c(y’)

y’ 2y’ y

ATC increases with y if the firm’s technology exhibits decreasing RTS

RETURNS-TO-SCALE
$
AND TOTAL COSTS
c(y)
c(2y’)
Slope = c(2y’)/2y’
c(y’) = AC(2y’).

Slope = c(y’)/y’
= AC(y’).

y’ 2y’ y

ATC decreases with y if the firm’s technology exhibits increasing RTS

RETURNS-TO-SCALE
$
AND TOTAL COSTS
c(y)
c(2y’) =2c(y’)
Slope = c(2y’)/2y’
= 2c(y’)/2y’
= c(y’)/y’
so AC(y’) = AC(2y’).
c(y’)

y’ 2y’ y

ATC is constant if the firm’s technology exhibits constant RTS

TYPES OF COST CURVES

• How are these cost curves related to each other?

• How are a firm’s long-run and short-run cost curves related?
AFC, AVC & ATC CURVES

• What does an AFC curve look like?

F
AFC( y ) =
y
• AFC(y) is a rectangular hyperbola so its graph looks like ...
$/output unit

AFC(y) → 0 as y → 

AFC(y)

0 y
AFC, AVC & ATC CURVES

• In a short-run with a fixed amount of at least one input, the Law of

Diminishing (Marginal) Returns must apply, causing the firm’s average
variable cost of production to increase eventually.
 Since AFC(y) → 0 as y → ,
$/output unit ATC(y) → AVC(y) as y → 

And since short-run AVC(y) must

eventually increase, ATC(y) must
eventually increase in a short-run.

ATC(y)

AVC(y)

AFC(y)

0 y
MARGINAL COST FUNCTION

• Marginal cost is the rate-of-change of variable production cost as the

output level changes. That is,

 cv ( y)
MC( y ) = .
y
MARGINAL COST FUNCTION

• The firm’s total cost function is

c( y ) = F + c v ( y )
• and the fixed cost F does not change with the output level y, so

 c v ( y )  c( y )
MC( y ) = = .
y y
• MC is the slope of both the variable cost and the total cost functions.
MARGINAL & VARIABLE COST FUNCTIONS

• Since MC(y) is the derivative of cv(y), cv(y) must be the integral of

MC(y). That is,
 cv ( y)
MC( y ) =
y
y
 c v ( y ) =  MC( z) dz.
0
MARGINAL & VARIABLE COST FUNCTIONS
$/output unit

y
c v ( y  ) =  MC( z) dz
0
MC(y)

VC of making y’ units

0 y y
MARGINAL & AVERAGE COST FUNCTIONS

• How is marginal cost related to average variable and average total

costs?
MARGINAL & AVERAGE COST FUNCTIONS

• The short-run MC curve intersects the short-run AVC curve from

below at the AVC curve’s minimum.
• Similarly, the short-run MC curve intersects the short-run ATC curve
from below at the ATC curve’s minimum.
$/output unit

MC(y)

ATC(y)

AVC(y)

y
SHORT-RUN & LONG-RUN TC CURVES

• A firm has a different short-run total cost curve for each possible short-
run circumstance.
• Suppose the firm can be in one of just three short-runs;
x2 = x2
or x2 = x2 x2 < x2 < x2.
or x2 = x2
SHORT-RUN & LONG-RUN TC CURVES

• In the long-run the firm is free to choose amongst these three since it
is free to select x2 equal to any of x2, x2, or x2.
• How does the firm make this choice?
 $
For 0  y  y, choose x2 = x2. cs(y;x2)

For y  y  y, choose x2 = x2.

For y < y, choose x2 = x2.

cs(y;x2)

cs(y;x2)
c(y), the
F firm’s long-
run total
cost curve.
F
F

y y
0 y
SHORT-RUN & LONG-RUN TC CURVES

• The firm’s long-run total cost curve consists of the lowest parts of the
short-run total cost curves. The long-run total cost curve is the lower
envelope of the short-run total cost curves.
SHORT-RUN & LONG-RUN TC CURVES

• If input 2 is available in continuous amounts then there is an infinity of

short-run total cost curves but the long-run total cost curve is still the
lower envelope of all of the short-run total cost curves.
 $
cs(y;x2)

cs(y;x2)

cs(y;x2) c(y)

F

F
F

0 y
SHORT-RUN & LONG-RUN ATC CURVES

• For any output level y, the long-run TC curve always gives the lowest
possible total production cost.
• Therefore, the long-run ATC curve must always give the lowest
possible average total production cost.
• The long-run ATC curve must be the lower envelope of all of the firm’s
short-run ATC curves.
SHORT-RUN & LONG-RUN ATC CURVES

• E.g. suppose again that the firm can be in one of just three short-runs;
x2 = x2
or x2 = x2 (x2 < x2 < x2)
or x2 = x2
then the firm’s three short-run average total cost curves are ...
$/output unit

ACs(y;x2)

ACs(y;x2)

ACs(y;x2)

y
SHORT-RUN & LONG-RUN ATC CURVES

• The firm’s long-run average total cost curve is the lower envelope of
the short-run average total cost curves ...
$/output unit

ACs(y;x2)

ACs(y;x2)

ACs(y;x2)

The long-run av. total cost

curve is the lower envelope AC(y)
of the short-run av. total cost curves.

y
SHORT-RUN & LONG-RUN MC CURVES

• Q: Is the long-run marginal cost curve the lower envelope of the firm’s
short-run marginal cost curves?
• A: No.
SHORT-RUN & LONG-RUN MC CURVES

• In the discrete case, the firm’s three short-run average total cost
curves are ...
$/output unit MCs(y;x2) MCs(y;x2)

ACs(y;x2)

ACs(y;x2)

ACs(y;x2)

MCs(y;x2)

MC(y), the long-run marginal

cost curve.

y
SHORT-RUN & LONG-RUN MC CURVES

• For any output level y > 0, the long-run marginal cost of production is
the marginal cost of production for the short-run chosen by the firm.
• This is always true, no matter how many and which short-run
circumstances exist for the firm.
• So for the continuous case, where x2 can be fixed at any value of zero
or more, the relationship between the long-run marginal cost and all of
the short-run marginal costs is ...
 SHORT-RUN & LONG-RUN MC CURVES
MC(y)
SRMCs
$/output unit

AC(y)

For each y > 0, the long-run MC equals the MC for the short-run chosen by the firm.