Lecture 7

Technology
 Technologies

A technology is a process by which

inputs are converted to an output
 E.g., labour, a computer, electricity,
software and cloud storage are being
combined to produce this lecture
 Technologies
 Usually,
several technologies will
produce the same product
 E.g.,
a whiteboard and pens can be
used instead of software
 Which technology is “best”?
 How do we compare technologies?
 Input Bundles

 xi denotes the amount used of input i

 An input bundle is a vector of the
input levels (x1, x2, … , xn)
 E.g. (x1, x2, x3) = (6, 0, 93)
 Production Functions

y denotes the output level

 The production function states the
maximum amount of output
technologically possible from an
input bundle

y  f ( x1 , , xn )
 Production Functions
One input, one output
Output Level y = f(x) is the
production
y’ function
y’ = f(x’) is the maximal
output level obtainable
from x’ input units

x’ x
Input Level
 Technology Sets

A production plan is an input bundle

and an output level (x1, … , xn, y)
A production plan is feasible if
y  f ( x1 , , xn )

 Thecollection of all feasible

production plans is the production set
 Production Sets
One input, one output
Output Level

Technically
y’ efficient plans
The production
y” Technically set
inefficient
plans
x’ x
Input Level
Technologies with Multiple Inputs
 What does a technology look like
when there is more than one input?
 The two-input case: A production
plan is a vector (x1,x2, y)
 Suppose the production function is

y  f ( x 1 , x 2 )  2x 1/ 3 1/ 3
1 x2 .
Technologies with Multiple Inputs
 E.g., the maximal output level possible
from the input bundle
(x1, x2) = (8, 1) is

1/3 1/3 1/3 1/3

y  2x 1 x 2  2  8  1  2  2  1  4.
 And the maximal output level possible
from (x1, x2) = (8, 8) is

1/3 1/3 1/3 1/3

y 2x 1 x 2  28 8  2  2  2  8.
Technologies with Multiple Inputs
Output, y

x2
(8,8)
(8,1)
x1
Technologies with Multiple Inputs
Technologies with Multiple Inputs

 Because representing two input

technologies in 3D is hard for the
human eye, we use instead a version
of level curves
 The y output unit isoquant is the set
of all input bundles that yield the
same output level y
Isoquants with Two Variable Inputs
x2

y 

y 
x1
Technologies with Multiple Inputs
 The complete collection of isoquants
is the isoquant map
 The isoquant map is a convenient
way to represent the production
function
1/3 1/3
 E.g. y  f(x 1, x 2 )  2x 1 x 2
Technologies with Multiple Inputs
x2

y
x1
Fixed-Proportions Technologies
A fixed-proportions (or Leontieff)
production function is of the form

y  min{ a 1 x 1 , a 2 x 2 ,  , a n x n }.

E.g. y  min{ x 1 , 2x 2 }
E.g.

with
Fixed-Proportions Technologies
x2 y  min{ x 1 , 2x 2 }

x1 = 2x2

7 min{x1,2x2 } = 14
4 min{x1,2x2 } = 8
2 min{x1,2x2 } = 4
4 8 14 x1
Perfect-Substitutes Technologies
A perfect-substitutes production
function is of the form

y  a 1 x 1  a 2x 2    a n x n .

E.g.E.g. y  x 1  3x 2

with
Perfect-Substitutes Technologies
y  x 1  3x 2
x2
x1 + 3x2 = 9
x1 + 3x2 = 18
x1 + 3x2 = 24
8
6
Isoquants are linear
3
and parallel
9 18 24 x1
 Marginal Products
y  f ( x1 , , xn )
 The marginal product of input i is the
rate-of-change of the output level as
the level of input i changes, holding
all other input levels fixed
 That is,
y
MPi 
 xi
 Marginal Products
E.g., if
1/ 3 2 / 3
y  f ( x1 , x 2 )  x1 x 2
then the marginal product of input 1 is
 Marginal Products
E.g., if
1/ 3 2 / 3
y  f ( x1 , x 2 )  x1 x 2
then the marginal product of input 1 is
 y 1  2/ 3 2/ 3
MP1   x1 x 2
 x1 3
 Marginal Products
E.g., if
1/ 3 2 / 3
y  f ( x1 , x 2 )  x1 x 2
then the marginal product of input 1 is
 y 1  2/ 3 2/ 3
MP1   x1 x 2
 x1 3
and the marginal product of input 2 is
 Marginal Products
E.g., if
1/ 3 2 / 3
y  f ( x1 , x 2 )  x1 x 2
then the marginal product of input 1 is
 y 1  2/ 3 2/ 3
MP1   x1 x 2
 x1 3
and the marginal product of input 2 is
 y 2 1/ 3  1/ 3
MP2   x1 x 2 .
 x2 3
 Marginal Products
Unless inputs are perfect substitutes, the
marginal product of one input depends
on the amount used of other inputs

E.g., MP1  1/3x 1 2/3 x 2/3

2

Then if x2 = 8, MP1  1/3x 1 2/3 8 2/3  4/3x 1 2/3

and if x2 = 27 then
 2/3 2/3  2/3
M P1  1/3x 1 27  3x 1 .
 Marginal Products

 The marginal product of input i is

diminishing if it becomes smaller as
the level of input i increases.
 That is, if
 MPi
 0.
 xi
 Marginal Products
1/ 3 2 / 3
E.g., if y  x 1 x 2 then
1  2/ 3 2/ 3 2 1/ 3  1/ 3
MP1  x1 x 2 and MP2  x1 x 2
3 3
 Marginal Products
1/ 3 2 / 3
E.g., if y  x 1 x 2 then
1  2/ 3 2/ 3 2 1/ 3  1/ 3
MP1  x1 x 2 and MP2  x1 x 2
3 3
so  MP1 2  5 / 3 2/ 3
  x1 x 2  0
 x1 9
 Marginal Products
1/ 3 2 / 3
E.g., if y  x 1 x 2 then
1  2/ 3 2/ 3 2 1/ 3  1/ 3
MP1  x1 x 2 and MP2  x1 x 2
3 3
so  MP1 2  5 / 3 2/ 3
  x1 x 2  0
 x1 9
and  MP
2   2 x1/ 3x  4 / 3  0 .
1 2
 x2 9
Both marginal products are diminishing
 Returns-to-Scale

 Marginal products describe the

change in output level as a single
input level changes
 Returns-to-scale describes how the
output level changes as all input
levels change in the same proportion
 E.g., all input levels doubled, or
halved
 Returns-to-Scale
If, for any input bundle (x1,…,xn),
f (kx 1 , kx 2 ,  , kx n )  kf ( x 1 , x 2 ,  , x n )
for any k > 1, then the technology described
by the production function f exhibits
constant returns-to-scale

E.g., If k = 2, then doubling all input levels

has doubled the output level
 Returns-to-Scale
One input, one output
Output Level
y = f(x)
2y’

Constant
y’ returns-to-scale

x’ 2x’ x
Input Level
 Returns-to-Scale
If, for any input bundle (x1,…,xn),

f (kx 1 , kx 2 ,  , kx n )  kf ( x 1 , x 2 ,  , x n )

For any k > 1 then the technology exhibits

diminishing returns-to-scale

E.g., If k = 2, then doubling all input levels

has less than doubled the output level
 Returns-to-Scale
One input, one output
Output Level

2f(x’) y = f(x)

f(2x’)
Decreasing
f(x’) returns-to-scale

x’ 2x’ x
Input Level
 Returns-to-Scale
If, for any input bundle (x1,…,xn),

f (kx 1 , kx 2 ,  , kx n )  kf ( x 1 , x 2 ,  , x n )

for any k > 1, then the technology exhibits

increasing returns-to-scale

E.g., If k = 2, then doubling all input levels

has more than doubled the output level
 Returns-to-Scale
One input, one output
Output Level
Increasing y = f(x)
returns-to-scale
f(2x’)

2f(x’)
f(x’)

x’ 2x’ x
Input Level
 Returns-to-Scale

A single technology can ‘locally’

exhibit different returns-to-scale
 Returns-to-Scale
One input, one output
Output Level

y = f(x)
Increasing
returns-to-scale
Decreasing
returns-to-scale

x
Input Level
 Examples of Returns-to-Scale
The Cobb-Douglas production function is
y  x 1a 1 x a2 2  x n
an
.
Scale up all input levels proportionately
by k. The output level becomes
a1 a2 an
(kx 1 ) (kx 2 )  (kx n )
 Examples of Returns-to-Scale
The Cobb-Douglas production function is
y  x 1a 1 x a2 2  x n
an
.
Scale up all input levels proportionately
by k. The output level becomes
a1 a2 an
(kx1 ) (kx 2 )  (kxn )
a1 a 2 an a1 a 2 an
 k k k x x  x
 Examples of Returns-to-Scale
The Cobb-Douglas production function is
y  x 1a 1 x a2 2  x n
an
.
Scale up all input levels proportionately
by k. The output level becomes
(kx1 ) a1 (kx 2 ) a 2 (kxn ) an
 k a1k a 2 k an x a1 x a 2 x an
 k a1  a 2  an x1a1 x a
2
2 x an
n
 Examples of Returns-to-Scale
The Cobb-Douglas production function is
y  x 1a 1 x a2 2  x n
an
.
Scale up all input levels proportionately
by k. The output level becomes
a1 a2 an
(kx1 ) (kx 2 ) (kxn )
a1 a 2 an a1 a 2 an
 k k k x x x
a1  a 2  an a1 a 2 an
k x1 x 2  xn
a1  an
k y.
 Examples of Returns-to-Scale
The Cobb-Douglas production function is
y  x 1a 1 x a2 2  x n
an
.

(kx 1 ) a 1 (kx 2 ) a 2  (kx n ) a n  k a 1    a n y .

The Cobb-Douglas technology’s returns-

to-scale depends on the sum of indices:
constant if a1+ … + an = 1
increasing if a1+ … + an > 1
decreasing if a1+ … + an < 1
(All this because the Cobb-Douglas function is homogeneous
of degree a1+ … + an )
 Returns-to-Scale

 Q: Can a technology exhibit

increasing returns-to-scale even if all
its marginal products are
diminishing?
 Returns-to-Scale

 Q: Can a technology exhibit

increasing returns-to-scale even if all
its marginal products are diminishing?
 A: Yes
 Why?
 Returns-to-Scale
A marginal product is the rate-of-
change of output as one input level
increases, holding all other input
levels fixed
 Marginal product diminishes because
the other input levels are fixed
 The new input’s units have each less
and less of other inputs to work with
 Returns-to-Scale
 When all input levels are increased
proportionately, new input units will
always have the same amount of
other inputs to work with
 Input productivities need not fall
 So returns-to-scale can be constant
or increasing
 Technical Rate of Substitution

 At
what rate can a firm substitute
one input for another without
changing its output level?
 Technical Rate of Substitution
x2

x'2

y

x'1 x1
 Technical Rate of Substitution
The slope of the isoquant is the
x2 rate at which input 2 must be
substituted by input 1 so as not
to change the output level.

x'2 It is the
Technical Rate of Substitution

y

x'1 x1
 Technical Rate of Substitution
 How is the TRS computed?
 Same as always
 Use total differentiation on y  f ( x 1 , x 2 ).
 We want to introduce a small change
(dx1, dx2) in the input bundle such that
the output does not change
y y
0 dx1  dx 2 .
 x1  x2
Technical Rate of Substitution
y y
0 dx1  dx 2
 x1  x2
rearranges to
y y
dx 1   dx 2
 x1  x2
so
dx 2  y/ x 1

dx 1  y/ x 2
 Technical Rate of Substitution
dx 2  y/ x 1 MP1
 
dx 1  y/ x 2 MP2

It is the rate at which input 2 must be

given up as input 1 increases to keep
the output level constant

It is the slope of the isoquant, as well as

the ratio of the marginal products
Technical Rate of Substitution; A
Cobb-Douglas Example
a b
y  f ( x1 , x 2 )  x1 x 2
MPs are
 y/ x 1  ax 1a 1x b2 an  y/ x 2  a b 1
bx 1 x 2 .
d
The TRS is

dx 2  y /  x1 ax1a  1xb2 ax 2
   .
dx1  y /  x2 a b1 bx1
bx1 x 2
Technical Rate of Substitution; A
Cobb-Douglas Example
x2 1/ 3 2 / 3 1 2
y  x1 x 2 ; a  and b 
3 3
ax 2 (1 / 3)x 2 x2
TRS    
bx1 ( 2 / 3 ) x1 2x 1

8 x2 8
TRS     1
2x 1 24

4 x1
Technical Rate of Substitution; A
Cobb-Douglas Example
x2 1/ 3 2 / 3 1 2
y  x1 x 2 ; a  and b 
3 3
ax 2 (1 / 3)x 2 x2
TRS    
bx1 ( 2 / 3 ) x1 2x 1
x2 6 1
TRS    
2x 1 2  12 4
6

12 x1
 Well-Behaved Technologies

Awell-behaved technology is
monotonic, and
convex
 Well-Behaved Technologies -
Monotonicity
 Monotonicity:More of any input
always generates more output

y y
monotonic
not
monotonic

x x
 Well-Behaved Technologies -
Convexity
 Convexity: If the input bundles x’
and x” both provide y units of output
then the mixture tx’ + (1-t)x”
provides at least y units of output,
for any 0 < t < 1
 Well-Behaved Technologies -
Convexity
x2

x'2

x"2
y
x'1 x"1 x1
 Well-Behaved Technologies -
Convexity
x2

x'2
 tx'1  ( 1  t ) x"1 , tx'2  ( 1  t ) x"2 
x"2
y
x'1 x"1 x1
 Well-Behaved Technologies -
Convexity
x2

x'2
 tx'1  ( 1  t ) x"1 , tx'2  ( 1  t ) x"2 
x"2 y
y
x'1 x"1 x1
 Well-Behaved Technologies -
Convexity
x2 Convexity in technology
implies that the TRS
x'2 decreases (in absolute value)
as x1 increases

Combining inputs
x"2 increases output
x'1 x"1 x1