Economic Applications of Implicit Differentiation
Economic Applications of Implicit Differentiation
1. Substitution of Inputs
Let Q = F(L , K ) be the production function in terms of labor and capital. Consider the isoquant
Q 0 = F(L , K ) of equal production. (This is the level curve of the function.) Thinking of K as a function
of L along the isoquant and using the chain rule, we get
0=
Q K
Q
+
.
L
K L
K
measures the rate of substitution of capital for labor needed to keep the output constant.
L
For the Cobb-Douglas production function Q = cL a K b . If the quantity produced Q is fixed and K is
considered a function of L, then differentiating
The derivative
0=
cbL a K b
L
= acL a1 K b + bcL a K b1
K
acL a1 K b
=
L
bcL a K b1
a K
.
=
b L
K
,
L
so
Thus, the marginal rate of technical substitution of capital for labor is a constant multiple of the average of
capital to labor.
2. Comparative-Static Analysis of Balancing Supply and Demand
Assume the supply and demand functions for a single commodity are
Q = S(P, T )
Q = D(P, Y )
(supply) and
(demand),
where Y is the income, T is the tax on the commodity, and P is the price. We do not assume a particular form
of the supply and demand functions, but we do assume that the supply increases with price and decreases
with tax while the demand decreases with price and increases with income: Assume that
S
> 0,
P
D
DP =
< 0,
P
S
< 0,
T
D
DY =
> 0.
Y
SP =
ST =
and
so
P
= DY
Y
P
(S P D P )
= ST ,
T
(S P D P )
and
and
P
DY
=
>0
Y
SP D P
P
ST
=
> 0.
T
SP D P
and
Thus, the price increases with both an increase in income and tax.
3. General Method of Implicit Differentiation
Following the books treatment of the general implicit function theorem, assume that there are m > 0
equations in n + m variables with n > 0,
F1 (x1 , . . . , xn , y1 , . . . , ym ) = C1
F2 (x1 , . . . , xn , y1 , . . . , ym ) = C2
..
..
.
.
Fm (x1 , . . . , xn , y1 , . . . , ym ) = Cm
such that at p = (a, b)
F1
F1
y (p) y (p)
m
..
..
det
6= 0.
.
.
Fm
Fm
(p)
(p)
y1
ym
By the implicit function theorem, there is a implicitly defined function y = h(x) such that C = F(x, h(x))
for all x near a. Thus, the (y1 , . . . , ym ) are the dependent variables and (x1 , . . . , xn ) are the independent
variables. Let H (x) = (x, h(x)), so C = F(H (x)). Differentiating this equation with respect to x and using
the chain rule, we get the matrix equation
I
0 = D F(p)D H (a) = Dx F(p), Dy F(p)
Dh(a)
= Dx F(p) + Dy F(p)Dh(a)
or
F1
F1
F1
y1
F1
y1
x (p) x (p) y (p) y (p) x (p) x (p)
0
m
1
1
1
n
n
..
..
..
..
..
.. .
(1) ... =
+
.
.
.
.
.
.
Fm
0
Fm Fm
Fm ym
ym
(p)
(p)
(p)
(p)
(p)
(p)
x1
xn
y1
ym
x1
xn
Solving for the partial derivatives of the dependent variables and taking the inverse of the square matrix
involving the partial derivatives of the Fi , we get
y1
x (p)
1
..
.
ym
(p)
x1
(2)
F1
y1
(p)
(p)
xn
y1
.. =
..
.
.
Fm
ym
(p)
(p)
xn
y1
1
F1
F1
(p)
(p)
ym
x1
..
..
.
.
Fm Fm
(p)
(p)
ym
x1
F1
(p)
xn
..
.
.
Fm
(p)
xn
In a given problem, we just write down the matrix equation (1) and solve them rather than use equation
(2). The reason is that it is easier to remember the correct form for the matrix equation (1): The rows
are determined by the coordinate functions (or equations) of the problem; the first matrix has the derivatives
with respect to the independent variables; the second matrix has the derivatives with respect to the dependent
variables and is multiplied by the unknown matrix of partial derivatives of the dependent variables with
yj
Fi
is multiplied by
as the
respect to the independent variables. Notice in the matrix product the term
yj
x`
chain rules says it must be.
= 0 and
= 0. Calculating these partial derivatives, we get
inputs that maximize profits satisfy
q1
q2
Q
=
p1 and
q1
q1
Q
0=
=
p2 .
q2
q2
0=
For the Cobb-Douglas production function Q = q13 q22 , these equations become
(3)
1 23 12
q q p1
3 1 2
1 1 1
0 = q13 q2 2 p2 .
2
0=
We have two equations and four variables p1 , p2 , q1 , and q2 . If we specified values for the prices, then we
would have two equations for the variables q1 and q2 which could determine their values.
To see that these two equations in (3) determine the amounts of inputs q1 and q2 that optimize profit in
terms of the prices p1 and p2 , we check the determinant of the matrix of partial derivatives with respect to
q1 and q2 :
-2 - 35 12 1 - 32 - 21
9 q1 q2 6 q1 q2
2
1
1 - 34 -1
- 43 -1
det
=
q
q
=
q1 q2 6= 0.
2
1
1
-3
2
1
1 - 3 - 2 -1 3 2
36
36
36
q q
q q
6 1 2
4 1 2
Therefore, the partial derivatives of q1 and q2 with respect to p1 and p2 satisfy the following matrix equation:
-2 - 53 12 1 - 23 - 21
q1 q1
q
q
q
q
9 1 2 6 1 2 p1 p2
0
-1 0
=
+
1 - 23 - 12 -1 13 -32 q2 q2
0
0 -1
q q
q q
p1 p2
6 1 2
4 1 2
q1
p1
q2
p1
-2 - 53 21 1 - 23 - 21 -1
q1
q1 q2
q1 q2
p2
= 9 2 1 6 1 -3
q2 1 - 3 - 2 -1 3 2
q q
q q
p2
6 1 2
4 1 2
-1 31 -32 -1 - 32 - 12
q
q
q
q
4
4 1 2
6 1 2
= 36 q13 q2
2
1
-1 - 3 - 2 -2 - 53 12
q q
q q
6 1 2
9 1 2
5
2
-1
1
3
3
2
2
-9
q
q
-6
q
q
1 2
1 2
,
=
2
-1
1
3
3
3
2
2
-6 q1 q2 -8 q1 q2
qi
are negative.
pj
Because the sum of the exponents in the production function is less than one, 31 + 12 < 1, the production
has decreasing return to scale. Although it is not obvious from the above treatment, this property caused the
partial derivatives to be negative and has the effect that the amounts of the inputs decreases with an increase
in either price.
The details of this example are complicated. However, the outline of the method is not complicated.
(a) First, some equations are derived (or given) relating several variables. In our example, (3) has two
equations with four variables. (b) Second, thinking of these equations as defining some variables in terms of
the others, we take partial derivatives. In our case, we take the partial derivatives with respect to p1 and p2 .
and all the partial derivatives
=
=
=
6= 0.
det a b =
y
x
xy
3 5 50
15
x y
Considering x and y as functions of a, b, p, and q, and differentiating the two equations with respect to
the four independent variables gives the following matrix equation:
# x x x x
"
p q a b p q
0
0
0
x y
=
+ a b
y y y y
0
ln(x) ln(y) 0 0
x y
a b p q
x
a
y
a
x
b
y
b
x
p
y
p
x
b -q
y
xy
q
0
0
x y
ln(x) ln(y) 0 0
y
p
pbx qay -a
x
q
#
"
xy
b
-q ln(x) -q ln(y) bx
y
=
qay pbx p ln(x) p ln(y) -a -ay
x
2
-x yq ln(y)
qay pbx
x yp ln(y)
qay pbx
-x yq ln(x)
qay pbx
=
x yp ln(x)
qay pbx
10050
3
50
3
3(25)(50) ln(5)
y
=
a
50
= 75 ln(5)
bx
qay pbx
-ax y
qay pbx
bx y
qay pbx
.
-ay 2 ln(y)
qay pbx
and
y
3(25)(50) ln(50)
=
b
50
= 75 ln(50).
The GDP is assume to be the sum of the consumption, investment, and government spending, Y = C +I +G.
The consumption is a function of after taxes income or the difference Y T , C = C(Y T ); the investment
is a function of the interest rate, I = I (r ); and the money supply is a function (called the liquidity function)
of GDP and the interest rate, M = L(Y, r ). The assumptions on the partial derivatives are that
L
> 0,
and
Y
C
Note that the partial derivative of C with respect to Y and T are
= C 0 and
Y
0 < C 0 (x) < 1,
I 0 (r ) < 0,
L
< 0.
r
C
= -C 0 .
T
We have obtained the following two functional relationships between the five variables:
(4)
0 = Y + C(Y T ) + I (r ) + G
0 = L(Y, r ) M
and
We consider G, M, and T as the independent variables which can be controlled; C and I are intermediate
variables (or functions); Y and r are dependent variables determined by G, M, and T through the variables
C and I .
Taking the partial derivatives of the two equations with respect to Y and r is an invertible matrix,
1 + C 0 I 0
= (C 0 1)L r I 0 L y > 0 :
1 = det
LY
Lr
The sign of 1 is positive because (C 0 1) < 0, L r < 0, (C 0 1) L r > 0, I 0 < 0, L Y > 0, and I 0 L Y > 0.
Therefore, these two equations define Y and r as functions of G, T , and M.
Taking the partial derivatives of the two equations with respect to G, T , and M (in that order), and
considering Y and r as functions of these variables, we get the following matrix equation:
Y
Y Y
0
0
0
0
1 -C 0
1 + C I G T M
=
+
0
0 0 -1
LY
L r r
r
r
.
G T M
Y Y
Y
1 Lr
-I 0
-1 C 0 0
G T M
=
r
r
r
1 -L Y C 0 1 0 0 1
.
G T M
1 -L r L r C 0
-I 0
=
.
1 L Y -L Y C 0 C 0 1
The partial derivatives of Y and r with respect to G are positive:
Y
-L r
r
LY
=
>0
and
=
> 0.
G
1
G
1
Thus, both the GDP, Y , and the interest rate, r , increase if government spending is increased while holding
the money supply and taxes fixed. The partial derivatives of Y and r with respect to T are negative, and both
Y and r decrease with increased taxes:
Y
Lr C 0
r
-L Y C 0
=
<0
and
=
< 0,
T
1
T
1
Finally,
Y
-I 0
r
C0 1
=
>0
and
=
< 0,
M
1
M
1
and Y increases and r decreases with an increased money supply.
This is the type of qualitative information that this type of analysis can give without knowing the particular
form of the functional dependence but only the signs of the rates of change.