Let
p
{\displaystyle p}
denote the price of output and
w
i
,
i
=
1
,
…
,
n
{\displaystyle w_{i},i=1,\ldots ,n}
denote the prices of
n
{\displaystyle n}
inputs. Let
x
i
{\displaystyle x_{i}}
be the amount of input
i
{\displaystyle i}
used in production and
y
{\displaystyle y}
be the output as determined by the production function, so
y
=
f
(
x
1
,
…
x
n
)
.
{\displaystyle y=f(x_{1},\ldots x_{n}).}
Profit as a function of the prices is derived by maximizing profit as a function of the prices and the quantity choices:
π
(
p
,
w
1
,
…
,
w
n
)
=
max
x
1
,
…
,
x
n
p
⋅
f
(
x
1
,
…
,
x
n
)
−
∑
i
=
1
n
w
i
⋅
x
i
{\displaystyle \pi (p,w_{1},\ldots ,w_{n})=\max _{x_{1},\ldots ,x_{n}}p\cdot f(x_{1},\ldots ,x_{n})-\sum _{i=1}^{n}w_{i}\cdot x_{i}}
Hotelling's Lemma says that if the profit function is differentiable and positive quantities of all inputs are used at the optimum, the profit-maximizing choices are:
y
∗
(
p
,
w
1
,
…
,
w
n
)
=
∂
π
(
p
,
w
1
,
…
,
w
n
)
∂
p
{\displaystyle y^{*}(p,w_{1},\ldots ,w_{n})={\frac {\partial \pi (p,w_{1},\ldots ,w_{n})}{\partial p}}}
x
i
∗
(
p
,
w
1
,
…
,
w
n
)
=
−
∂
π
(
p
,
w
1
,
…
,
w
n
)
∂
w
i
{\displaystyle x_{i}^{*}(p,w_{1},\ldots ,w_{n})=-{\frac {\partial \pi (p,w_{1},\ldots ,w_{n})}{\partial w_{i}}}}
Proof of Hotelling's lemma
edit
The lemma uses the same reasoning as the envelope theorem .
The function for maximum profit can be written as
π
(
p
,
w
1
,
…
,
w
n
,
x
1
∗
,
…
,
x
n
∗
)
=
p
⋅
f
(
x
1
∗
,
…
,
x
n
∗
)
−
∑
i
=
1
n
w
i
⋅
x
i
∗
,
{\displaystyle \pi (p,w_{1},\ldots ,w_{n},x_{1}^{*},\ldots ,x_{n}^{*})=p\cdot f(x_{1}^{*},\ldots ,x_{n}^{*})-\sum _{i=1}^{n}w_{i}\cdot x_{i}^{*},}
where
x
1
∗
,
…
,
x
n
∗
{\displaystyle x_{1}^{*},\ldots ,x_{n}^{*}}
are the maximizing inputs corresponding to the optimal output
y
∗
=
f
(
x
1
∗
,
…
,
x
n
∗
)
{\displaystyle y^{*}=f(x_{1}^{*},\ldots ,x_{n}^{*})}
. Because the inputs are maximizing profit, the first order conditions hold:
∂
π
∂
x
i
|
x
i
=
x
i
∗
=
p
∂
f
∂
x
i
|
x
i
=
x
i
∗
−
w
i
=
0.
{\displaystyle {\frac {\partial \pi }{\partial x_{i}}}{\bigg |}_{x_{i}=x_{i}^{*}}=p{\frac {\partial f}{\partial x_{i}}}{\bigg |}_{x_{i}=x_{i}^{*}}-w_{i}=0.}
(1 )
Taking the derivative of profit with respect to
p
{\displaystyle p}
at the optimal values of the inputs yields
d
π
d
p
=
∑
j
=
1
n
∂
π
∂
x
j
|
x
j
=
x
j
∗
∂
x
j
∂
p
+
∂
π
∂
p
=
∂
π
∂
p
=
f
(
x
1
∗
,
…
,
x
n
∗
)
=
y
∗
(
p
,
w
1
,
…
,
w
n
)
,
{\displaystyle {\frac {d\pi }{dp}}=\sum _{j=1}^{n}{\frac {\partial \pi }{\partial x_{j}}}{\bigg |}_{x_{j}=x_{j}^{*}}{\frac {\partial x_{j}}{\partial p}}+{\frac {\partial \pi }{\partial p}}={\frac {\partial \pi }{\partial p}}=f(x_{1}^{*},\ldots ,x_{n}^{*})=y^{*}(p,w_{1},\ldots ,w_{n}),}
where
∂
π
∂
x
j
|
x
j
=
x
j
∗
=
0
{\displaystyle {\frac {\partial \pi }{\partial x_{j}}}{\bigg |}_{x_{j}=x_{j}^{*}}=0}
for every input
j
{\displaystyle j}
because of (1 ).
Similarly, taking the derivative with respect to input price
w
i
{\displaystyle w_{i}}
yields
d
π
d
w
i
=
∑
j
=
1
n
∂
π
∂
x
j
|
x
j
=
x
j
∗
∂
x
j
∂
w
j
+
∂
π
∂
w
i
=
∂
π
∂
w
i
=
−
x
i
∗
{\displaystyle {\frac {d\pi }{dw_{i}}}=\sum _{j=1}^{n}{\frac {\partial \pi }{\partial x_{j}}}{\bigg |}_{x_{j}=x_{j}^{*}}{\frac {\partial x_{j}}{\partial w_{j}}}+{\frac {\partial \pi }{\partial w_{i}}}={\frac {\partial \pi }{\partial w_{i}}}=-x_{i}^{*}}
QED