Division of the Humanities
and Social Sciences
Euler’s Theorem for Homogeneous Functions
KC Border
October 2000
v. 2017.10.27::16.34
1 Definition Let X be a subset of Rn . A function f : X → R is homoge-
neous of degree k if for all x ∈ X and all λ > 0 with λx ∈ X,
f (λx) = λk f (x).
• Note that if 0 ∈ X and f is homogeneous of degree k ̸= 0, then
f (0) = f (λ0) = λk f (0), so setting λ = 2, we see f (0) = 2k f (0), which
implies f (0) = 0.
• A constant function is homogeneous of degree 0.
• If a function is homogeneous of degree 0, then it is constant on rays
from the the origin.
• Linear functions are homogenous of degree one.
• The economists’ favorite homogeneous function is the weighted geo-
metric mean with domain Rn+ , which they know as the Cobb–Douglas
function [3],
f (x1 , . . . , xn ) = xα1 1 xα2 2 · · · xαnn ,
where each αi > 0. It is homogeneous of degree α = α1 + · · · + αn .
• If f1 , . . . , fm are homogeneous
(
of degree )k and g is homogeneous of
degree ℓ, then h(x) = g f1 (x), . . . , fm (x) is homogeneous of degree
k + ℓ.
2 Remark Usually the domain of f will be taken to be a cone, that is, a
set in Rn closed under multiplication by positive scalars. Indeed, we often
take the domain to be the set Rn+ of vectors with nonnegative components,
or the set Rn++ of vectors with strictly positive components. Apostol [2,
Exercises 8, 9, p. 287] proves the following theorem for any open domain.
The proof is the same as the one used here.
1
KC Border Euler’s Theorem for Homogeneous Functions 2
3 Euler’s theorem Let f : Rn+ → R be continuous and also differentiable
on Rn++ . Then f is homogeneous of degree k if and only if for all x ∈ Rn++ ,
∑
n
Di f (x)xi = kf (x). (*)
i=1
Proof : (=⇒) Assume f is homogeneous of degree k. Let x ∈ Rn++ . Define
the function g : [0, ∞) → R (depending on x) by
g(λ) = f (λx) − λk f (x),
and note that for all λ > 0,
g(λ) = 0.
Therefore
g ′ (λ) = 0
for all λ > 0. But by the chain rule,
∑
n
g ′ (λ) = Di f (λx)xi − kλk−1 f (x).
i=1
Evaluate this at λ = 1 to obtain (∗).
(⇐=) Suppose
∑
n
kf (x) = Di f (x)xi
i=1
for all x ∈ Rn++ . Fix any such x and again define g : I → R (depending on
x) by
g(λ) = f (λx) − λk f (x)
and note that g(1) = 0. Then for λ > 0,
∑
n
g ′ (λ) = Di f (λx)xi − kλk−1 f (x)
i=1
( n )
∑
−1
= λ Di f (λx)λxi − kλk−1 f (x)
i=1
= λ−1 kf (λx) − kλk−1 f (x),
so
( )
λg ′ (λ) = k f (λx) − λk f (x)
= kg(λ).
v. 2017.10.27::16.34
KC Border Euler’s Theorem for Homogeneous Functions 3
Since λ is arbitrary, g satisfies the following differential equation:
k
g ′ (λ) − g(λ) = 0
λ
and the initial condition g(1) = 0. By Theorem 7 below,
∫ λ
g(λ) = 0 · eA(λ) + e−A(λ) 0 · eA(t) dt = 0
1
∫
where, irrelevantly, A(λ) = − 1λ kt dt = −k ln λ. This implies g is identi-
cally zero, so f is homogeneous on Rn++ . Continuity guarantees that f is
homogeneous on X.
4 Corollary Let f : Rn+ → R be continuous and differentiable on Rn++ . If
f is homogeneous of degree k, then Dj f (x) is homogeneous of degree k − 1.
Proof if f is twice differentiable: By the first half of Euler’s theorem,
∑
n
Di f (x)xi = kf (x)
i=1
so differentiating both sides with respect to the j th variable,
( n )
∑
Dj Di f (x)xi = kDj f (x)
i=1
or
∑
n
Dij f (x)xi + Dj f (x) = kDj f (x)
i=1
or
∑
n
Dij f (x)xi = (k − 1)Dj f (x). (1)
i=1
Thus Dj f (x) is homogeneous of degree (k − 1) by second half of Euler’s
theorem.
Proof without twice differentiability: The difference quotients satisfy
f (λx + λh) − f (λx) λk f (x + h) − λk f (x) f (x + h) − f (x)
= = λk−1
∥λh∥ λ∥h∥ ∥h∥
whenever λ > 0. Thus f is differentiable at λx if and only if it is differen-
tiable at x and Di f (λx) = λk−1 Di f (x) for all i = 1, . . . , n.
v. 2017.10.27::16.34
KC Border Euler’s Theorem for Homogeneous Functions 4
5 Corollary If f is homogeneous of degree k, then
Di f (λx) Di f (x)
=
Dj f (λx) Dj f (x)
for λ > 0 and x ∈ Rn++ .
Proof : By Corollary 4 each fi satisfies Di f (λx) = λk−1 Di f (x), so
Di f (λx) λk−1 Di f (x) Di f (x)
= k−1 = .
Dj f (λx) λ Dj f (x) Dj f (x)
6 Corollary If f is homogeneous of degree 1 and twice differentiable, then
the Hessian matrix [Dij f (x)] is singular for all x ∈ Rn++ .
Proof : By (1),
∑
n
Dij f (x)xi = (k − 1)Dj f (x).
i=1
When k = 1 this becomes [Dij f (x)]x = 0 in matrix terms, so for x ̸= 0 we
conclude that [Dij f (x)] is singular.
7 Theorem (Solution of first order linear differential equations)
Assume P, Q are continuous on the open interval I. Let a ∈ I, b ∈ R.
Then there is one and only one function y = f (x) that satisfies the initial
value problem
y ′ + P (x)y = Q(x)
with f (a) = b. It is given by
∫ x
−A(x) −A(x)
f (x) = be +e Q(t)eA(t) dt
a
where ∫ x
A(x) = P (t) dt.
a
For a proof see [1, Theorems 8.2 and 8.3, pp. 309–310].
v. 2017.10.27::16.34
KC Border Euler’s Theorem for Homogeneous Functions 5
References
[1] Apostol, T. M. 1967. Calculus, 2d. ed., volume 1. Waltham, Mas-
sachusetts: Blaisdell.
[2] Apostol, T. M. 1969. Calculus, 2d. ed., volume 2. Waltham, Mas-
sachusetts: Blaisdell.
[3] Cobb, C. W. and P. H. Douglas. 1928. A theory of production. American
Economic Review 18(1):139–165. Supplement, Papers and Proceedings
of the Fortieth Annual Meeting of the American Economic Association.
v. 2017.10.27::16.34