Production Functions

Abhishek Dureja

Teaching Fellow: Shreya Kapoor

Plaksha University
 Production Functions
Introduction

▶ Till now we have seen how individuals make consumption and labor
supply decisions
▶ We now turn to how firms make choices and take decisions
▶ The principal activity of any firm is to turn inputs into outputs
▶ Economists are interested in the choices the firm makes in
accomplishing this goal
▶ To avoid discussing many of the engineering intricacies involved
Economists construct an abstract model of production
▶ Economists model the relationship between inputs and outputs
by using a production function
▶ Production functions represent the production technology that
converts inputs into output

2 / 83
 Production Functions
Introduction

▶ The production function Economists use can take a general form

q = f (k, l, ....) (1)

where q represents the firm’s output of a particular good during a

period
k represents the machine (i.e., capital) usage during the period
l represents hours of labor input
▶ There remains a possibility of other variables affecting the
production process
▶ The general production function (1) provides any conceivable set
of inputs, the engineer’s solution to the problem of how best
to combine those inputs to get output

3 / 83
 Production Functions
Introduction

Production Function
▶ The firm’s production function for a particular good, q,

q = f (k, l) (2)

shows the maximum amount of the good that can be produced

using alternative combinations of capital (k) and labor (l)
▶ This means that as we keep changing the levels of k and l, the
level of output produced will keep changing
▶ Importantly, a change in output brought about by a change in
one of the productive inputs while keeping the other
productive input as fixed
▶ This leads us to the idea of marginal product

(For convenience, we are limiting to capital (k) and labor (l) as the
only two productive inputs in the production function)
4 / 83
 Production Functions
Marginal Physical Product

Marginal physical product

▶ The marginal physical product of an input is the additional
output that can be produced by using one more unit of that
input while holding all other inputs constant
▶ Mathematically,

∂q
Marginal physical product of capital = MPk = = fk
∂k

▶ If the amount of capital input increases by 1 unit, keeping

labor input as fixed, what is the increase in the output
∂q
Marginal physical product of labor = MPl = = fl
∂l

▶ If the amount of labor input increases by 1 unit, keeping

capital input as fixed, what is the increase in the output
5 / 83
 Production Functions
Marginal Physical Product

▶ Mathematical definitions of marginal product use partial derivatives,

=⇒ All other input usage is held constant while the input of

interest varies
▶ For example, consider a farmer hiring one more laborer to harvest
the crop but holding all other inputs constant
=⇒ The extra output this laborer produces is that laborer’s
marginal physical product
▶ Assume 50 workers on a farm are able to produce 100 Kgs of wheat
per year
→ Whereas 51 workers, with the same land and equipment, can
produce 102 Kgs
=⇒ The marginal physical product of the 51st worker is 2 Kgs per
year

6 / 83
 Production Functions
Diminishing Marginal Productivity

▶ Marginal physical product of an input depends on how much

of that input is used

▶ For example, labor cannot be added indefinitely to a given field

(while keeping the amount of equipment, fertilizer, and so forth
fixed)
=⇒ Adding more labor will eventually exhibit some
deterioration in its productivity

▶ This assumption is the assumption of diminishing marginal

physical productivity

▶ Mathematically, it is an assumption about the second-order partial

derivatives of the production function

7 / 83
 Production Functions
Diminishing Marginal Productivity

▶ Mathematically, this means

∂MPk ∂2f
= = fkk = f11 < 0
∂k ∂k

=⇒ As we keep adding more capital input, while holding labor

input fixed, the total output keeps increasing but at a
decreasing rate

∂MPl ∂2f
= = fll = f22 < 0
∂l ∂l

=⇒ As we keep adding more labor input, while holding capital

input fixed, the total output keeps increasing but at a
decreasing rate

8 / 83
 Production Functions
Diminishing Marginal Productivity

▶ The assumption of diminishing marginal productivity was originally

proposed by the nineteenth-century economist Thomas Malthus
▶ Malthus worried that rapid increases in population would result
in lower labor productivity
▶ His gloomy predictions for the future of humanity led economics to
be called the dismal science
▶ But the mathematics of the production function suggests that such
gloom may be misplaced
▶ Changes in the marginal productivity of labor over time
depend not only on how labor input is growing but also on
changes in other inputs, such as capital
=⇒ We must also be concerned with MPlk

9 / 83
 Production Functions
Diminishing Marginal Productivity

▶ In most cases, MPlk = flk > 0

=⇒ Marginal product of labor will increase if capital increases

( =⇒ Marginal product of capital will increase if labor increases)
=⇒ Declining labor productivity as both l and k increase is not a
foregone conclusion

▶ Indeed, labor productivity has risen significantly since Malthus’ time

▶ Because increases in capital inputs (along with technical

improvements) have offset the impact of decreasing marginal
productivity

10 / 83
 Production Functions
Average Physical Productivity

▶ Economists are often interested in measuring the productivity of

labor (employees)
▶ How can labor productivity be measured?
▶ Labor productivity usually means average productivity
▶ Average productivity is easily measured (say, x Kgs of wheat per
labor-hour input) and it is often used as a measure of efficiency
▶ Average product of labor (APl ) is

Output q f (k, l)
APl = = =
Labor input l l

▶ Notice that APl also depends on the level of capital used

11 / 83
 Production Functions
Example: A Two-Input Production Function

q = f (k, l) = 600k 2 l 2 − k 3 l 3

▶ Assume k = 10

=⇒ q = f (k, l) = 60000l 2 − 1000l 3

▶ Marginal product of labor is given by:

∂q
MPl = = 120000l − 3000l 2
∂l
▶ MPl diminishes as l increases, and eventually becoming negative

=⇒ q reaches a maximum value

12 / 83
 Production Functions
Example: A Two-Input Production Function

▶ Finding where the MPl is maximum, set MPl = 0

120000l − 3000l 2 = 0

=⇒ 40l = l 2 =⇒ l = 40

▶ At l = 40, MPl is maximum

13 / 83
 Production Functions
Example: A Two-Input Production Function

q
APl = = 60000l − 1000l 2
l

▶ APl is an inverted parabola that reaches its maximum value when

∂APl
= 60000 − 2000l = 0
∂l

=⇒ APl is maximum at l = 30

▶ At l = 30 =⇒ APl = 900000 and MPl = 900000

▶ When APl is at a maximum, average and marginal
productivities of labor are equal

14 / 83
 Production Functions
Maximum of APl

q
APl =
l

Taking derivative wrt l (and using division rule, we get)

∂APl lMPl − q
=
∂l l2

∂APl l × MPl − q
For maxima = =0
∂l l2

=⇒ l × MPl = q =⇒ MPl = APl

▶ Therefore, whenever APl is maximum, APl = MPl

15 / 83
 Production Functions
Isoquant Maps and the Rate of Technical Substitution

▶ Assume a production function of the form:

f (k, l)

▶ Isoquant: An isoquant shows those combinations of k and l

that can produce a given level of output (say, q0 )
▶ Mathematically, an isoquant records the set of k and l that satisfies

f (k, l) = q0

▶ An isoquant indicates the possible substitution of one input for

another in a production function
▶ An isoquant (from iso, meaning equal) records those combinations
of k and l that are able to produce a given quantity of output

16 / 83
 Production Functions
Isoquant Maps and the Rate of Technical Substitution

▶ All those combinations of k and l that fall on the curve labeled

q = 10 are capable of producing 10 units of output per period
▶ We can use either (lA , kA ) or (lB , kB ) to produce 10 units of output

17 / 83
 Production Functions
Isoquant Maps and the Rate of Technical Substitution

▶ As was the case for indifference curves, there are infinitely many
isoquants in the k–l plane

▶ Each isoquant represents a different level of output

▶ Isoquants record successively higher levels of output as we move in a

North-Easterly direction

▶ Using more of each of the inputs will permit output to increase

▶ Two other isoquants (for q = 20 and q = 30) are also shown in the
figure

18 / 83
 Production Functions
Isoquant Maps and the Rate of Technical Substitution

19 / 83
 Production Functions
Isoquant Maps and the Rate of Technical Substitution

▶ Isoquant map is similar to an Indifference map because both

represent contour maps of a particular function

▶ For isoquants, however, the labeling of the curves is measurable —

An output of 10 units per period has a quantifiable meaning
=⇒ Economists are more interested in studying the shape of
production functions than in examining the exact shape of utility
functions

20 / 83
 Production Functions
The Marginal Rate of Technical Substitution (RTS)

▶ The slope of an isoquant shows how one input can be traded

for another while holding output constant
▶ Examining the slope provides information about the technical
possibility of substituting labor (l) for capital (k)
Marginal rate of technical substitution
▶ The marginal rate of technical substitution (RTS) shows the
rate at which labor (l) can be substituted for capital (k) while
holding output constant along an isoquant
▶ Mathematically,
dk
RTS (l for k) = − q=q0
dl

▶ The notation indicates that output is to be held constant as l is

substituted for k

21 / 83
 Production Functions
The Marginal Rate of Technical Substitution (RTS)

dk
RTS (l for k) = − q=q0
dl

▶ The particular value of this trade-off rate (RTS) will depend on

i The level of output

ii Quantities of capital and labor being used

=⇒ RTS depends on the point on the isoquant map at which

the slope is to be measured

22 / 83
 Production Functions
RTS and Marginal Productivities

▶ While we are on an isoquant, the output produced (say, q0 ) remains

the same

▶ We substitute one input for the other

→ If we increase l =⇒ k decreases (so that output remains

constant)

▶ We can say that along an isoquant, the quantity of k depends on

the quantity of l

▶ Mathematically, k is a function of l

=⇒ k can be generally be written as k(l)

23 / 83
 Production Functions
RTS and Marginal Productivities

▶ Therefore the production function for a given isoquant can be

written as:
q0 = f (k(l), l)

▶ Taking total derivative

∂f ∂f
dk + dl = 0
∂k ∂l

=⇒ fk dk + fl dl = 0

dk fl
=⇒ − =
dl fk

dk MPl
=⇒ RTS = − q=q0
=
dl MPk
24 / 83
 Production Functions
RTS and marginal productivities

MPl
RTS =
MPk

▶ Isoquants are negatively sloped (because if l increases =⇒ k

decreases)

▶ Both MPl and MPk are non-negative (as no firm would choose to
use a costly input that reduces output)
=⇒ The RTS will be positive

25 / 83
 Production Functions
Reasons for a diminishing RTS

▶ The isoquants are convex to the origin

▶ Along an isoquant, RTS is diminishing

▶ For high ratios of k to l, the RTS is a large positive number

→ More capital can be given up to increase labor input by one more

unit

▶ However, when a lot of labor is already being used, the RTS is low

▶ =⇒ Only a small amount of capital can be traded for an additional

unit of labor

▶ The convexity of isoquants imposes assumptions on the

marginal products

26 / 83
 Production Functions
Reasons for a diminishing RTS

▶ Lets compute dMRTS

dl

dMRTS d( ffkl )
=
dl dl

▶ For a given isoquant, quantities of k and l are interdependent

→ Also fl and fk are functions of both k and l

dMRTS fk (fll + flk dk dk

dl ) − fl (fkl + fkk dl )
=
dl (fk )2

▶ Using the fact that dk

dl = − ffkl and fkl = flk

27 / 83
 Production Functions

dMRTS fk (fll + flk dk dk

dl ) − fl (fkl + fkk dl )
=
dl (fk )2

▶ Using the fact that dk

dl = − ffkl and fkl = flk

dRTS f 2 fll − 2fk fl fkl + fl 2 fkk

= k
dl (fk )3

▶ Since fk > 0 =⇒ Denominator is positive

▶ For diminishing RTS, the numerator must be negative
▶ Because fll and fkk are both assumed to be negative, the numerator
will be negative if fkl is positive

28 / 83
 Production Functions
Importance of Cross-Productivity Effects

▶ fkl represents the cross-productivity effects

▶ If fkl > 0 =⇒ dRTS

dl <0
→ fkl > 0 =⇒ If workers had more capital, they would have higher
marginal productivities

▶ Alternatively, RTS can still be diminishing if fk2 fll + fl 2 fkk < 2fk fl fkl

▶ While assuming diminishing RTS, we are assuming

i Diminishing marginal productivities for each input (fkk < 0; fll < 0)

ii Marginal productivities diminish rapidly enough to compensate for

any possible negative cross-productivity effects (if we do not want to
impose strong assumption of fkl > 0)

29 / 83
 Production Functions
Example: Diminshing RTS

q = f (k, l) = 600k 2 l 2 − k 3 l 3

▶ General marginal productivity functions for this production function

are
∂q
MPl = fl = = 1200k 2 l − 3k 3 l 2
∂l

∂q
MPk = fk = = 1200kl 2 − 3k 2 l 3
∂k

▶ Each of these marginal productivities depend on the values of both

inputs (k and l)
▶ Marginal productivities of k and l will be positive for values of k and
l for which kl < 400

30 / 83
 Production Functions
Example: Diminshing RTS

q = f (k, l) = 600k 2 l 2 − k 3 l 3

fll = 1200k 2 − 6k 3 l

fkk = 1200l 2 − 6kl 3

▶ This production function exhibits diminishing marginal productivities

for sufficiently large values of k and l
▶ fkk < 0 and fll < 0 if kl > 200
▶ However, even within the range 200 < kl < 400 where the marginal
productivity relations for this function behave normally, this
production function may not necessarily have a diminishing
RTS
31 / 83
 Production Functions
Example: Diminshing RTS

▶ The reason is the cross-marginal productivity

▶ Cross-differentiation of either of the marginal productivity gives the
cross-marginal productivity

fkl = flk = 2400kl − 9k 2 l 2

▶ The cross-marginal productivity is positive only for kl < 266

=⇒ RTS will be declining for 200 < kl < 266

▶ For cases where 266 < kl < 400,

→ The diminishing marginal productivities exhibited by the function

are sufficient to overcome the influence of a negative value for fkl on
the convexity of isoquants

32 / 83
 Production Functions
Returns to Scale

▶ Given a production function, Economists want to characterize the

production function

▶ One question which is of interest about the production function is:

→ How output responds to increases in all inputs together

▶ For example, suppose that all inputs were doubled:

→ Would output double? Less than double? More than

double?

▶ This is a question of the returns to scale exhibited by the

production function

33 / 83
 Production Functions
Returns to Scale

▶ This question of returns to scale has been of interest to economists

ever since Adam Smith intensively studied the production of
pins
▶ Smith identified two forces that may come into operation when the
conceptual experiment of doubling all inputs was performed
▶ First, a doubling of scale permits a greater division of labor and
specialization of function
=⇒ Efficiency might increase
=⇒ Production might more than double
▶ Second, doubling of inputs may also entail some loss in efficiency

→ Because managerial overseeing may become more difficult

given the larger scale of the firm
▶ Which of these two effect will have a greater effect is an important
empirical question
34 / 83
 Production Functions
Returns to Scale

Returns to scale

▶ If the production function is given by

q = f (k, l)

→ If all inputs are multiplied by the same positive constant t (where

t > 1), then we classify the returns to scale of the production as

i Constant Returns to Scale if f (tk, tl) = tf (k, l) = tq

ii Decreasing Returns to Scale if f (tk, tl) < tf (k, l) = tq

iii Increasing Returns to Scale if f (tk, tl) > tf (k, l) = tq

35 / 83
 Production Functions
Returns to Scale

▶ If a proportionate increase in inputs increases output by the same

proportion
=⇒ The production function exhibits constant returns to
scale

▶ If output increases less than the proportionate increase in inputs

=⇒ The function exhibits diminishing returns to scale

▶ If output increases more than the proportionate increase in inputs

=⇒ There are increasing returns to scale

36 / 83
 Production Functions
Returns to Scale

▶ It is theoretically possible for a function to exhibit constant

returns to scale for some levels of input usage and increasing
or decreasing returns for other levels
▶ Economists examine returns to scale of a production function while
considering only a fairly narrow range of variation in input usage and
the related output level

37 / 83
 Production Functions
Constant Returns to Scale

▶ Constant returns to scale is a theoretical construct

▶ Is there an economic rationale for production functions to

exhibit constant returns to scale ?

▶ There are economic reasons why a firm’s production function might

exhibit constant returns to scale

▶ If the firm operates many identical plants

=⇒ It may increase or decrease production simply by varying

the number of plants in operation
=⇒ The firm can double output by doubling the number of
plants it operates
→ This will require employing precisely twice as many inputs

38 / 83
 Production Functions
Constant Returns to Scale

▶ Empirical studies of production functions often find that

returns to scale are roughly constant for the firms that were
studied
→ This holds true for outputs close to the firms’ established
operating levels
→ The firms may exhibit increasing returns to scale as they expand
to their established size

▶ Since constant returns to scale is an empirical fact

=⇒ It is worthwhile to examine these production functions more

39 / 83
 Production Functions
Constant Returns to Scale

▶ Since constant returns to scale means

f (tk, tl) = t 1 f (k, l) = tq

=⇒ The production function is homogeneous of degree 1

▶ If a function is homogeneous of degree k

=⇒ Its derivatives are homogeneous of degree k − 1

▶ Since a production function is homogeneous of degree 1

=⇒ Its partial derivative with respect to k and l will be

homogeneous of degree 0 (zero)
=⇒ MPl and MPk are homogeneous of degree 0 (zero)

40 / 83
 Production Functions
Constant Returns to Scale

▶ MPl and MPk are homogeneous of degree 0 (zero)

∂f (k,l)
▶ Since MPk = ∂k

∂f (tk, tl) ∂f (k, l)

=⇒ = = MPk for any t > 0
∂k ∂k

∂f (k,l)
▶ Similarly, MPl = ∂l

∂f (tk, tl) ∂f (k, l)

=⇒ = = MPl for any t > 0
∂l ∂l

41 / 83
 Production Functions
Constant Returns to Scale

▶ Since for any t > 0

∂f (tk, tl) ∂f (k, l)

MPk = =
∂k ∂k

∂f (tk, tl) ∂f (k, l)

MPl = =
∂l ∂l

▶ Let t = 1
l

∂f ( kl , 1)
=⇒ MPk =
∂k

∂f ( kl , 1)
=⇒ MPl =
∂l

42 / 83
 Production Functions
Constant Returns to Scale

∂f ( kl , 1)
MPk =
∂k

∂f ( kl , 1)
MPl =
∂l

▶ The marginal productivity of any input depends only on the

ratio of capital to labor input, not on the absolute levels of
these inputs

▶ This helps in explaining differences in productivity among industries

or across countries

43 / 83
 Production Functions
Constant Returns to Scale

▶ For instance: Industries with high capital-labor ratio may have

higher productivity of labor
=⇒ Higher income levels
=⇒ Richer (Developed) countries

▶ Indeed developed countries have higher capital-labor ( kl ) ratios

▶ Many growth models also take capital-labor ( kl ) ratio as a key factor

in explaining growth rates
→ As capital-labor ratio is a key indicator of an economy’s
productivity level

44 / 83
 Production Functions
Homothetic production functions

▶ Since MPl and MPk depend on the ratio of k

l

▶ Also, RTS = MPl

MPk

=⇒ RTS for any constant returns-to-scale production function

will depend only on the ratio of the inputs, not on their
absolute levels
k
=⇒ RTS depends upon the ratio of l (capital-labor ratio) and the
absolute level of k and l
=⇒ The production function is Homothetic
=⇒ Isoquants will be radial expansions of one another
=⇒ Along any ray through the origin (where the ratio k/l
does not change) the slopes of successively higher isoquants
are identical

45 / 83
 Production Functions
Homothetic production functions

▶ Since RTS depends only on the ratio of k

l

=⇒ Isoquants will be radial expansions of one another

=⇒ Along any ray through the origin (where the ratio k/l
does not change) the slopes of successively higher isoquants
are identical

▶ This property of the isoquant map is very useful

▶ The substitution possibilities between k and l depend upon on the

ratio of kl

46 / 83
 Production Functions
Homothetic production functions

47 / 83
 Production Functions
Homothetic production functions

▶ Assume a bread order of 200 loaves can be completed in one day by

3 bakers working with 3 ovens or by 2 bakers working with 4 ovens
=⇒ RTS of ovens for bakers is one-for-one
=⇒ One extra oven can be substituted for one baker
▶ Assume this production process exhibits constant returns to
scale
=⇒ Two large bread orders of 400 loaves (total) can be completed
in one day, either by 6 bakers with 6 ovens or by 4 bakers with 8
ovens
=⇒ Two ovens are substituted for two bakers, so again the RTS is
one-for-one
▶ In constant returns-to-scale cases

=⇒ Expanding the level of production does not alter

trade-offs among inputs
48 / 83
 Production Functions
Homothetic production functions

▶ In constant returns-to-scale cases

=⇒ Expanding the level of production does not alter trade-offs

among inputs

=⇒ If production functions are homothetic, RTS depends only

on the ratio of kl

49 / 83
 Production Functions
Homothetic production functions

▶ If a production function is constant returns to scale

=⇒ A monotonic transformation can change the returns to scale of

the production

▶ But the monotonic transformation cannot change the RTS of

the production function

▶ Assume f (k, l) is a constant returns to scale production function

▶ Let us take a monotonic transformation: F (k, l) = [f (k, l)]γ

where γ is any positive exponent

50 / 83
 Production Functions
Homothetic production functions

F (k, l) = [f (k, l)]γ

▶ Assuming γ > 1

=⇒ F (tk, tl) = [f (tk, tl)]γ = [tf (k, l)]γ

=⇒ F (tk, tl) = t γ [f (k, l)]γ > t[f (k, l)]γ

for any t > 1

▶ The exponent γ captures the degree of the increasing returns to
scale
▶ A doubling of inputs would lead to a four-fold (eight-fold) increase
in output if γ = 2 (γ = 3)

51 / 83
 Production Functions
Homothetic production functions

▶ Therefore for γ > 1

=⇒ Monotonically transformed production function, F (k, l)

displays increasing returns to scale

▶ Similarly, for γ < 1, the monotonically transformed production

function, F (k, l) displays decreasing returns to scale

▶ But the monotonically transformed production function, F (k, l),

continues to be homothetic
=⇒ RTS of monotonically transformed production function,
F (k, l), depends on the kl ratio

52 / 83
 Production Functions
The n-input case

▶ The definition of returns to scale can be easily generalized to a

production function with n inputs
▶ If the production function is given by

q = f (x1 , x2 , ...., xn )

and if all inputs are increased t-times, where t > 1, then

f (tx1 , tx2 , ...., txn ) = t k f (x1 , x2 , ...., xn )

i If k > 1 =⇒ The production function displays increasing returns

to scale

ii If k = 1 =⇒ The production function displays constant returns

to scale

iii If k < 1 =⇒ The production function displays decreasing

returns to scale
53 / 83
 Production Functions
The n-input case

▶ It is important to note that the requirement is that

→ All inputs be increased by the same proportion: t

▶ In real-world production processes, this provision may make little
economic sense
▶ For example, a firm may have only one manager, and the number of
managers need not necessarily be doubled even if all other inputs
were
▶ Alternatively – if the output of a farm depends on the fertility of the
soil
=⇒ It may not be literally possible to double the cultivated area
while maintaining the same level of fertility
→ Because the new land may not be as good as that already under
cultivation

54 / 83
 Production Functions
The n-input case

▶ =⇒ Hence some inputs may have to be fixed (or at least

imperfectly variable) for most practical purposes

▶ In such cases, some degree of diminishing productivity (a result of

increasing employment of variable inputs) seems likely

▶ But this cannot be called diminishing returns to scale because of the

presence of some inputs that are held fixed

55 / 83
 Production Functions
Elasticity of Substitution

▶ An important characteristic of the production function is how easy

it is to substitute one input for another

▶ This question relates to the shape of a single isoquant (rather than

the isoquant map)

▶ Along an isoquant

→ The rate of technical substitution decreases

k
→ The capital-labor ratio also decreases (i.e. l ↓)

▶ We now wish to define a parameter that measures this degree

of responsiveness

56 / 83
 Production Functions
Elasticity of Substitution

▶ If the RTS does not change at all for changes in k

l

=⇒ Substitution is easy
→ Because the ratio of the marginal productivities of the two
inputs does not change as the input mix changes

▶ Alternatively, if the RTS changes rapidly for small changes in k

l

=⇒ Substitution is difficult
→ Because minor variations in the input mix has a substantial
effect on the inputs’ relative productivities

57 / 83
 Production Functions
Elasticity of Substitution

▶ We want a measure to the responsiveness of changes in RTS to

changes in kl ratio

▶ A scale-free measure of this responsiveness is provided by the

elasticity of substitution

▶ The elasticity of substitution (for small changes) is given by

d ln( kl ) d ( kl ) RTS
σ= = × k
d lnRTS d RTS (l )

58 / 83
 Production Functions
Elasticity of Substitution

▶ Alternatively, for discrete changes in k and l, the elasticity of

substitution is given by

percent ∆ ( kl ) ∆ ( k ) ∆ RTS
σ= = kl ×
percent ∆ RTS (l ) RTS

∆ ( kl ) RTS
=⇒ σ = × k
∆ RTS (l )

▶ We will use the following to denote the elasticity of substitution

d ( kl ) RTS
σ= × k
d RTS (l )

59 / 83
 Production Functions
Elasticity of Substitution

Elasticity of substitution
▶ For the production function

q = f (k, l)

The elasticity of substitution (σ) measures the proportionate change

in kl relative to the proportionate change in the RTS along an
isoquant

percent ∆ ( kl ) d ( kl ) RTS d ln( kl ) d ln( kl )

σ= = × k = =
percent ∆ RTS d RTS (l ) d lnRTS d ln( ffkl )

60 / 83
 Production Functions
Elasticity of Substitution

▶ Is the elasticity of substitution positive or negative?

▶ Elasticity of substitution is always positive

k
→ Because along an isoquant l and RTS move in the same
direction
=⇒ The value of σ is always positive

61 / 83
 Production Functions
Elasticity of Substitution: Graphical Explanation

62 / 83
 Production Functions
Elasticity of Substitution: Graphical Explanation

▶ As we move from point A to point B on the isoquant

k
=⇒ Both the RTS and the ratio l will change

▶ The RTS changes from RTSA to RTSB

k
▶ The l changes from ( kl )A to ( kl )B

▶ We are interested in the relative magnitude of these change

63 / 83
 Production Functions
Elasticity of Substitution: Graphical Explanation

▶ If σ is high
k
=⇒ The RTS will not change much relative to l

=⇒ The isoquant will be close to linear

▶ If σ value is low

=⇒ Isoquant is sharply curved

k
=⇒ The RTS will change by a substantial amount as l changes
▶ It is convenient to assume that σ is constant along an isoquant

▶ If the production function is also homothetic

=⇒ All the isoquants are merely radial blowups and σ will be

the same along all isoquants

64 / 83
 Production Functions
Examples of Production Function

1. Linear (Perfect Substitutes) Production Function

▶ The Linear production function is given by

q = f (k, l) = αk + βl

▶ All isoquants for this production function are parallel straight

β
lines with slope − α
β
=⇒ RTS =
α

▶ This production function exhibits constant returns to scale

f (tk, tl) = α(tk) + β(tl) = t(αk + βl) = tf (k, l)

=⇒ f(tk,tl) = t f(k,l)

65 / 83
 Production Functions
Examples of Production Function: Linear Production Function

66 / 83
 Production Functions
Examples of Production Function

▶ Since RTS is constant along the isoquant

=⇒ Change in RTS along the isoquant is zero

=⇒ Elasticity of substitution is infinite for linear production
function

▶ Linear production implies perfect substitution between the

inputs (k and l)

67 / 83
 Production Functions
Examples of Production Function

2. Fixed proportions (Perfect Complements) Production Function

▶ The fixed proportions production function is given by

q = min(αk, βl) where α > 0, β > 0

▶ The isoquants for this production function are L-shaped

▶ This production implies that the inputs are to be combined in
fixed-proportions to produce the output
▶ A firm would always operate at the corner of an isoquant
▶ Operating anywhere else is inefficient

→ Because the same output could be produced with fewer inputs by

moving along the isoquant toward the corner

68 / 83
 Production Functions
Examples of Production Function: Linear Production Function

69 / 83
 Production Functions
Examples of Production Function: Fixed Proportions Production Function

q = min(αk, βl) where α > 0, β > 0

▶ At the vertex of any isoquant, αk = βl =⇒ k β

l = α

▶ This is how the inputs k and l are to be combined

▶ Since the inputs are to be combined in fixed proportion to produce

the output
=⇒ It is not possible to substitute one input for the other
=⇒ Elasticity of Substitution for this production function is
zero

70 / 83
 Production Functions
Examples of Production Function: Fixed Proportions Production Function

▶ Since the curve is L-shaped, there are only two RTS ratios

▶ One RTS ratio is represented by the vertical line (of the L-shaped)
of isoquant
▶ The other RTS ratio is represented by the horizontal line (of the
L-shaped) of isoquant
▶ As we move from the vertical segment to the horizontal segment

=⇒ RTS ratio changes

=⇒ RTS changes from infinity to zero
=⇒ A large change in the RTS as we move from vertical to
horizontal segment
=⇒ Elasticity of substitution is Zero

71 / 83
 Production Functions
Examples of Production Function: Fixed Proportions Production Function

q = min(αk, βl) where α > 0, β > 0

▶ Suppose that αk < βl, then q = αk

=⇒ Capital is the binding constraint in this production process

=⇒ The employment of more labor would not increase output
=⇒ Marginal product of labor is zero
=⇒ Additional labor is superfluous in this case

72 / 83
 Production Functions
Examples of Production Function: Fixed Proportions Production Function

▶ Similarly, if αk > βl,

=⇒ Labor is the binding constraint on output

=⇒ Additional capital is superfluous

▶ When αk = βl

=⇒ Both inputs are fully utilized

=⇒ Production takes place at a vertex on the isoquant map

▶ If both inputs are costly, this is the only cost-minimizing place

to operate

▶ The locus of all such vertices is a straight line through the origin
β
with a slope given by α

73 / 83
 Production Functions
Examples of Production Function: Cobb-Douglas Production Function

3. Cobb-Douglas Production Function

▶ Cobb–Douglas production function, provides a middle ground
between the two polar cases previously discussed

▶ Cobb–Douglas production function is given by

q = f (k, l) = Ak α l β

where A, α, and β are all positive constants

▶ Isoquants for the Cobb–Douglas production function are

convex to the origin

74 / 83
 Production Functions
Examples of Production Function: Cobb-Douglas Production Function

▶ The elasticity of substitution for these production functions is 1

(one)
k k
RTS = =⇒ ln RTS = ln
l l

k
=⇒ artialln RTS = artialln
l

∂ln kl
=⇒ =1
∂ln RTS

=⇒ σ = 1

75 / 83
 Production Functions
Examples of Production Function: Cobb-Douglas Production Function

▶ The Cobb–Douglas function can exhibit any degree of returns to

scale, depending on the values of α and β

f (tk, tl) = A(tk)α (tl)β = t α+β Ak α l β

=⇒ f (tk, tl) = t α+β f (k, l)

▶ If α + β = 1 =⇒ Constant returns to scale

▶ If α + β < 1 =⇒ Decreasing returns to scale

▶ If α + β > 1 =⇒ Increasing returns to scale

76 / 83
 Production Functions
Examples of Production Function: Cobb-Douglas Production Function

▶ The Cobb–Douglas function has also proved to be useful in many

applications because it is linear in logarithms

ln q = ln A + α ln k + β ln l

▶ α is the elasticity of output with respect to capital input

▶ β is the elasticity of output with respect to labor input

▶ These constants can be estimated from actual data and are used to
measure returns to scale (by using sum α + β)

77 / 83
 Production Functions
Examples of Production Function: CES production function

4. Constant Elasticity of Substitution (CES) production function

▶ A functional form that incorporates all three previous cases is the
constant elasticity of substitution (CES) production function
▶ CES production function is of the form
γ
q = f (k, l) = [k ρ + l ρ ] ρ

where ρ <= 1, ρ ̸= 1, and γ > 0

▶ CES production function has a constant elasticity of
substitution
▶ CES production function allows σ to take on any value (including
the σ of past three production function)

78 / 83
 Production Functions
Examples of Production Function: CES production function

γ
q = f (k, l) = [k ρ + l ρ ] ρ

▶ For γ > 1 the function exhibits increasing returns to scale

▶ For γ < 1 the function exhibits decreasing returns to scale

▶ The elasticity of substitution of this production function is given by

1
σ=
1−ρ

79 / 83
 Production Functions
Examples of Production Function: CES production function

▶ CES production function boils down to Linear production

function when ρ = 1
=⇒ σ = ∞

▶ CES production function boils down to Fixed proportions

production function when ρ = −∞
=⇒ σ = 0

▶ CES production function boils down to Cobb-Douglas

production function when ρ = 0
=⇒ σ = 1

80 / 83
 Production Functions
Examples of Production Function: CES production function

▶ Prooving that the CES production function boils down to different

production functions for different values of ρ requires using limits
argument

▶ For example:

γ
lim f (k, l) = lim [k ρ + l ρ ] ρ = k + l
ρ→1 ρ→1

81 / 83
 Production Functions
Examples of Production Function: CES production function

▶ Often the CES production function is used with a distributional

weight, α(0 ≤ α ≤ 1) to indicate the relative significance of the
inputs
γ
=⇒ q = f (k, l) = [αk ρ + (1 − α)l ρ ] ρ

▶ With constant returns to scale (γ = 1) and ρ = 0, this function

converges to the Cobb–Douglas form

q = f (k, l) = k α l 1−α

82 / 83
 References

Chapter 9
Snyder, Christopher and Nicholson, Walter. (2012). Microeconomic
Theory: Basic Principles and Extensions

83 / 83