2
$\begingroup$

I am studying homothetic functions using Mathematics for Economists by Simon and Blume. I am reading their proof of the following theorem:

Theorem$\quad$ Let $u: \mathbb{R}_+^\mathbf{n} \to \mathbb{R}$ be a strictly monotonic function. Then, $u$ is homothetic if and only if for all $\mathbf{x}$ and $\mathbf{y}$ in $\mathbb{R}_+^\mathbf{n}$, \begin{align} u(\mathbf{x}) \geq u(\mathbf{y}) \iff u(\alpha \mathbf{x}) \geq u(\alpha \mathbb{y}),\quad \textit{for all}\quad \alpha > 0.\tag1 \end{align}

I feel that there is a lot of typo and misleading notation in their proof, and that the proof in the textbook is not complete. So I rewrite the proof. I would like to know if my proof is rigorous and complete. I would really appreciate it if someone could help me check!

Here is my attempt:

Proof$\quad$ We first show that if $u$ satisfies (1), it is homothetic. Let $\mathbf{e}$ denote the vector $(1, 1, \dots, 1)$, that spans the diagonal $\Delta$ in $\mathbb{R}^{\mathbf{n}}$. Define function $f: \mathbb{R}_+ \to \mathbb{R}$ by \begin{align*} f(t) = u(t\mathbf{e}). \end{align*} Since $u$ is strictly increasing, so is $f$; and therefore, $f$ has a strictly increasing inverse $f^{-1}$. Then \begin{align*} f \circ (f^{-1} \circ u) = (f \circ f^{-1}) \circ u = u. \end{align*} To prove that $u = f \circ (f^{-1} \circ u)$ is homothetic, we need only show that $(f^{-1} \circ u)$ is homogeneous.

$\quad$ For any scalar $a$, the function $a \mapsto f^{-1}(a)$ tells how far up the diagonal $\Delta$ the level set $u^{-1}(a)$ meets $\Delta$. Consequently, $f^{-1}(u(\mathbf{x}))$ tells how far up $\Delta$ the $u$-level set through $\mathbf{x}$ crosses $\Delta$. Analytically, $t = f^{-1}(u(\mathbf{x}))$ is the solution of \begin{align} u(\mathbf{x}) = u(t \mathbf{e}).\tag2 \end{align} Let $\alpha > 0$ be a scalar. By (1), \begin{align} u(\mathbf{x}) = u(t \mathbf{e}) \implies u(\alpha \mathbf{x}) = u(\alpha t \mathbf{e}).\tag3 \end{align} But (3) indicates that $s = \alpha t = \alpha f^{-1}(u(\mathbf{x}))$ is the solution of \begin{align} u(\mathbf{\alpha x}) = u(s \mathbf{e}).\tag4 \end{align} Since (2) indicates that $s = f^{-1}(u(\alpha \mathbf{x}))$ is also the solution of (4), we have that \begin{align*} f^{-1}(u(\alpha \mathbf{x})) = \alpha f^{-1}(u(\mathbf{x})); \end{align*} thus, $(f^{-1} \circ u)$ is homogeneous of degree one. Since $(f^{-1} \circ u)$ is homogeneous and $f$ is increasing, $u = f \circ (f^{-1} \circ u)$ is homothetic.

$\quad$ To prove the converse, suppose first that $u$ is linear homogeneous, that is homogeneous of degree one, and that $\alpha > 0$. These two properties yield \begin{align*} u(\mathbf{x}) \geq u(\mathbf{y}) & \iff \alpha u(\mathbf{x}) \geq \alpha u(\mathbf{y})\\ & \iff u(\alpha \mathbf{x}) \geq u(\alpha \mathbf{y}) \end{align*} so, property (1) holds.

$\quad$ More generally, suppose that $u$ is homothetic, so that $u = g_1 \circ v$, with $g_1$ increasing and $v$ homogeneous of degree $k$. Write $v$ as $g_2 \circ h$, where $g_2(z) = z^k$ and $h(\mathbf{x}) = v(\mathbf{x})^\frac{1}{k}$. We check that $h$ is homogeneous of degree one: \begin{align*} h(\alpha \mathbf{x}) & = v(\alpha \mathbf{x})^\frac{1}{k}\\ & = \left(\alpha^k v(\mathbf{x})\right)^\frac{1}{k}\\ & = \alpha (v(\mathbf{x}))^\frac{1}{k}\\ & = \alpha h(\mathbf{x}). \end{align*} We also have that $g_2$ is increasing. Thus, we can write $u$ as $u = p \circ h$ with $p \equiv g_1 \circ g_2$ increasing and $h$ linear homogeneous.

$\quad$ Once again, suppose $\alpha > 0$. Since $p$ is strictly increasing, it has a strictly increasing inverse $p^{-1}$. Therefore, \begin{align*} u(\mathbf{x}) \geq u(\mathbf{y}) & \iff p^{-1}(u(\mathbf{x})) \geq p^{-1}(u(\mathbf{y}))\\ & \iff h(\mathbf{x}) \geq h(\mathbf{y})\\ & \iff \alpha h(\mathbf{x}) \geq \alpha h(\mathbf{y})\\ & \iff h(\alpha \mathbf{x}) \geq h(\alpha \mathbf{y})\\ & \iff p(h(\alpha \mathbf{x})) \geq p(h(\alpha \mathbf{y}))\\ & \iff u(\alpha \mathbf{x}) \geq u(\alpha \mathbf{y}) \end{align*} and so $u$ satisfies property (1).

Thanks a lot for any help!

$\endgroup$

1 Answer 1

2
$\begingroup$

Looks rigorous and complete to me. And it is indeed clearer than the corresponding proof in Simon and Blume.

$\endgroup$

Your Answer

By clicking “Post Your Answer”, you agree to our terms of service and acknowledge you have read our privacy policy.

Not the answer you're looking for? Browse other questions tagged or ask your own question.