Production Theory

Dr. Manuel Salas-Velasco

University of Granada, Spain

Page 1 Dr. Manuel Salas-Velasco

The Production Function
• Production refers to the transformation of inputs into outputs
(or products)
• An input is a resource that a firm uses in its production process
for the purpose of creating a good or service
• A production function indicates the highest output (Q) that a
firm can produce for every specified combinations of inputs
(physical relationship between inputs and output), while holding
technology constant at some predetermined state
• Mathematically, we represent a firm’s production function as:

Q = f (L, K)
Assuming that the firm produces only one type of output with two inputs,
labor (L) and capital (K)

Page 2 Dr. Manuel Salas-Velasco

The Production Function
Q = f (L, K)

• The quantity of output is a function of, or depend on, the

quantity of labor and capital used in production
• Output refers to the number of units of the commodity
produced
• Labor refers to the number of workers employed
• Capital refers to the amount of the equipment used in
production
• We assume that all units of L and K are homogeneous or
identical
• Technology is assumed to remain constant during the period of
the analysis

Page 3 Dr. Manuel Salas-Velasco

 Production Theory

The Production Function in the

Short Run

Page 4 Dr. Manuel Salas-Velasco

The Short Run

• The short run is a time period in which the quantity

of some inputs, called fixed factors, cannot be
increased. So, it does not correspond to a specific
number of months or years
• A fixed factor is usually an element of capital (such
as plant and equipment). Therefore, in our
production function capital is taken to be the fixed
factor and labor the variable one

Q  f (L, K )

Page 5 Dr. Manuel Salas-Velasco

Total, Average and Marginal Products
• Total product (TP) is the total amount that is
produced during a given period of time
• Total product will change as more or less of the
variable factor is used in conjunction with the given
amount of the fixed factor
• Average product (AP) is the total product divided by
the number of units of the variable factor used to
produce it
• Marginal product (MP) is the change in total product
resulting from the use of one additional unit of the
variable factor
Page 6 Dr. Manuel Salas-Velasco
Output with Fixed Capital and Variable
Labor
Quantity of labor (L) Total product (TP) Marginal product (MP) Average product (AP)

0 0
1 50 50 50.00
2 110 60 55.00
3 390 280 130.00
4 520 130 130.00
5 580 60 116.00
6 630 50 105.00
7 650 20 92.86
8 650 0 81.25
9 640 -10 71.11

Page 7 Dr. Manuel Salas-Velasco

Total Product, Average Product and
Marginal Product Curves

700 300
280
Total product (units per time period)

600 TP 260

AP and MP (units per time period)

240
220
500
200
180
400 160
140
300 120
100 APL
200 80
60
40
100
20 MPL
0
0 -20 0 1 2 3 4 5 6 7 8 9 10
0 1 2 3 4 5 6 7 8 9 -40
Quantity of labor per time period, L
Quantity of labor per time period, L

Page 8 Dr. Manuel Salas-Velasco

Relationships among Total, Marginal an
Average Products of Labor
C
TP B • With labor time continuously
Total
product divisible, we can smooth TP, MPL
and APL curves

A • The TP curve increases at an

A = inflection point increasing rate up to point A;
past this point, the TP curve
rises at a decreasing rate up to
Labor point C (and declines thereafter)
APL
A • The MPL rises up to point A,
MPL
becomes zero at C, and is
Marginal product B negative thereafter
• The APL raises up to point B
Average product
and declines thereafter (but
remains positive as long TP is
C Labor positive)
LA LB LC
Page 9 Dr. Manuel Salas-Velasco
Law of Diminishing Returns
C • This law states that as
TP B
Total additional units of an input are
product
used in a production process,
while holding all other inputs
A constant, the resulting
increments to output (or total
product) begin to diminish
beyond some point (after A, in
Labor the bottom graph)
APL
MPL A • As the firm uses more and more
units of the variable input with the
Marginal product B same amount of the fixed input,
each additional unit of the variable
input has less and less of the fixed
Average product input to work and, after this point,
the marginal product of the variable
C Labor input declines
LA LB LC
Page 10 Dr. Manuel Salas-Velasco
Stages of Production
C
B The relationship between the MPL
TP
Total and APL curves can be used to
product define three stages of production
of labor (the variable input)
A Stage I of labor
Is the range of production for which
increases in the use of a variable input
cause increases in its average product

Labor Stage II of labor

APL Is the range for which increases in the

MPL A use of a variable input causes decreases
in its average product, while values of its
B associated marginal product remain
Marginal product
nonnegative

Stage III of labor

Average product Is the range for which the use of a
variable input corresponds to negative
C Labor
values for its marginal product
3 4 8
Page 11 Dr. Manuel Salas-Velasco
In Terms of Calculus …
• Marginal product of labor = change in output/change in labor
input =
• If we assume that inputs and outputs are continuously or
infinitesimally divisible (rather than being measured in discrete
units), then the marginal product of labor would be:
Q
MPL 
L Q  f (L, K )
Example. Let’s consider the following production function:

Q = 8 K1/2 L1/2 Cobb-Douglas production function

1
TPL  8 L 2

1  12
MPL  8K 2 1
L If K = 1   12
2
MPL  4 L
Page 12 Dr. Manuel Salas-Velasco
The Production or Output Elasticity of Labor

The elasticity of output with respect to the labor

input measures the percentage change in output for a 1
percent change in the labor input, holding the capital
input constant

Q L
EL 
L Q

L
EL  MPL
Q

Page 13 Dr. Manuel Salas-Velasco

Production Theory

The Production Function in the

Long Run

Page 14 Dr. Manuel Salas-Velasco

The Long Run

• The long run is a time period in which all inputs may be varied
but in which the basic technology of production cannot be
changed
• The long run corresponds to a situation that the firm faces
when is planning to go into business (to expand the scale of its
operations)
• Like the short run, the long run does not correspond to a
specific length of time
• We can express the production function in the form:

Q  f (L, K)

Page 15 Dr. Manuel Salas-Velasco

Production Isoquants
K An isoquant is a set of input combinations that can
be used to produce a given level of output

This curve indicates that a firm can produce the

Units per time period

K1 a specified level of output from input combinations

(L1, K1), (L2, K2), (L3, K3), …

As we move down from one point on an

isoquant to another, we are substituting
K2 b one factor for another while holding
output constant

K3 c
Q  f (L, K)

L
L1 L2 L3
Units per time period
Page 16 Dr. Manuel Salas-Velasco
Marginal Rate of Technical Substitution
• The marginal rate of technical substitution (MRTS)
K measures the rate at which one factor is substituted for
another with output being held constant

• Since we measure K on the vertical axis, the MRTS

represents the amount of capital that must be
Units per time period

K1 a sacrificed in order to use more labor in the production

process, while producing the same level of output:
ΔK/ΔL (the slope of the isoquant which is negative)

K2 b • We multiply the ratio by -1 in

order to express the MRTS as a
positive number
K3 c Q  f (L, K)
K
MRTS  
L L
L1 L2 L3
Units per time period
Page 17 Dr. Manuel Salas-Velasco
MRTS When Labor and Capital Are
Continuously Divisible
Q  f (L, K)

Let’s take the total differential of the production function:

Q Q
dQ  dL  dK
L K
Setting this total differential equal zero (since output does not change along a
given isoquant):
Q
Q Q Q Q dK L dK MPL
0 dL  dK  dK  dL  
L K K L d L  Q dL MPK
K

d K MPL
  MRTS 
MPL
d L MPK MPK

In production theory, the marginal rate of technical substitution is equal to the

ratio of marginal products (in consumer theory, the marginal rate of substitution is
equal to the ratio of marginal utilities)

Page 18 Dr. Manuel Salas-Velasco

A Numerical Example
Assume the production function is:
1 2
Q3K L 3 3

The marginal product functions are:

Q Q 2  23
 13
MPK  3L3
1
MPL  3K
1
3 2
L K
L 3
K 3

1 1
MPL 2K3L 3 2K
MRTS    23 2 
MPK K L 3 L
If we specify the level of output as Q = 9 units, and the firm uses 3 units of labor, then
the amount of capital used is:
1 2 2 1
9  3 K 3 3 3  3  3 3 K 3  K  3 units

2 3
MRTS  2 This result indicates that the firm can substitute 2 units of capital for 1
3 unit of labor and still produce 9 units of output

Page 19 Dr. Manuel Salas-Velasco

Production Theory

Econometric Analysis of
Production Theory

Page 20 Dr. Manuel Salas-Velasco

The Cobb-Douglas Production Function

Q  A L1 K 2 eu
• Q = output
• L = labor input
• K = labor capital
• u = stochastic disturbance term
• e = base of natural logarithm
• The parameter A measures, roughly speaking, the scale of
production: how much output we would get if we used one unit
of each input
• The parameters beta measure how the amount of output
responds to changes in the inputs
Page 21 Dr. Manuel Salas-Velasco
The Cobb-Douglas Production Function

Q  A L1 K 2 eu
• Our problem is to obtain an estimated function:

ˆ ˆ1
Q AL K ˆ2

• However, if we take logarithms in both sides, we have:

lnQ  lnA  1 lnL   2 lnK  u

β0

This is the log-log, double-log or log-linear model

Page 22 Dr. Manuel Salas-Velasco

The Properties of the Cobb-Douglas
Production Function

lnQ  0  1 lnL  2 lnK  u

Fitting this equation by the method of least squares, we have:

1. The estimated coefficient β1 is the elasticity of output with

respect to the labor input; that is, it measures the percentage
change in output for a 1 percent change in the labor input,
holding the capital input constant
2. Likewise, the estimated coefficient β2 is the elasticity of output
with respect to the capital input, holding the labor input
constant

Page 23 Dr. Manuel Salas-Velasco

The Properties of the Cobb-Douglas
Production Function
lnQ  0  1 lnL  2 lnK  u
3. The sum of the estimated coefficients β1 and β2
gives information about the returns to scale
• If this sum is 1, then there are constant returns to
scale; that is, doubling the inputs will double the output,
tripling the inputs will triple the output, and so on
• If the sum is less than 1, there are decreasing returns to
scale; e.g. doubling the inputs will less than double the
output
• If the sum is greater than 1, there are increasing returns
to scale; e.g. doubling the inputs will more than double
the output
Page 24 Dr. Manuel Salas-Velasco
Cobb-Douglas Production Function: The
Agricultural Sector in Taiwan (1958-1972)
• The log-linear model:

ln Yi  0  2 ln X 2i  3 X 3i  ui

• Regression by the OLS method: d = 0.891

ln Yˆi   3.338  1.499 ln X 2i  0.490 ln X 3i

** **
output elasticity output elasticity
of labor of capital

1.989
** Significant at 5%-level increasing returns to scale

Page 25 Dr. Manuel Salas-Velasco

Cobb-Douglas Production Function: The
Agricultural Sector in Taiwan (1958-1972)
Evidence of No autocorrelation Evidence of
positive negative
autocorrelation autocorrelation
Zone of Zone of
indecision indecision
0 dL dU 2 4 – dU 4 – dL 4
d = 0.891 d = 1.939
0.946 1.543 2.457 3.054
0.814 1.750 2.250 3.186

Regression including a time variable: d = 1.939

ln Yˆi  9.348  0.878 ln X 2i  0.469 ln X 3i  0.064 TIME

** * **
** Significant at 5%-level
* Significant at 10%-level

Page 26 Dr. Manuel Salas-Velasco

Cobb-Douglas Production Function: The
U.S. Bell Companies (1981-82)
• The log-linear model:

ln Y j   0   1 ln K j   2 ln M j  3 ln L j  error

• Regression by the OLS method: d = 1.954 (no autocorrelation)

ln Yˆj  1.461  0.401 ln K j  0.373 ln M j  0.202 ln L j

** ** ** **
output elasticity
of labor
A 1 percent increase in the labor input
led on the average to about a 0.2
percent increase in the output
** Significant at 5%-level

Page 27 Dr. Manuel Salas-Velasco

Problems in Regression Analysis
• Regression analysis may face two main econometric problems
when we use cross-sectional data (data on economic units for a
given year or other time period):
– Multicollinearity
– Heteroscedasticity
• This arises when the assumption that the variance of the error term is
constant for all values of the independent variables is violated
• This situation leads to biased standard errors
• When the pattern of errors or residuals points to the existence of
heteroscedasticity, the researcher may overcome the problem by using
logs or by running a weighted least squares regression
• Nowadays, several computer packages (STATA, LIMDEP, …) present
robust standard errors (using White’s procedure)

Page 28 Dr. Manuel Salas-Velasco

Detection of Heteroscedasticity: The
Breusch-Pagan Test
• Step 1. Estimate the model using OLS and obtain the residuals, ûi
• Step 2. Obtain the maximum likelihood estimator of σ2: ∑ûi2/n
• Step 3. Construct the variable pi: divide the squared residuals obtained
from regression (ûi2) by the number obtained in step 2
• Step 4. Regress pi on the X’s and obtain ESS (explained sum of
squares)
• Step 5. Obtain the B-P statistic (ESS/2) and compare it with the critical
chi-square value

B-P B-P

There is not heteroscedasticity There is heteroscedasticity

critical chi-
square value
Page 29 Dr. Manuel Salas-Velasco