# Concave production function implies convex cost function

Let's assume we have an increasing production function $f:\mathbb{R^+} \to \mathbb{R^+}$

Now, assume this production function is concave and that the price of input z is fixed (this is a single-input and single output case). I want to show this implies the corresponding cost function $C^f(w,q)$ is convex.

My thoughts:

let $z,z' \in \mathbb{R^+}$ where W.L.O.G $z'>z$ and let $f(z)=q$ and $f(z')=q$

Since $f$ is concave, we take $\alpha \in [0,1]$ $s.t$:

$$f(\alpha z +(1- \alpha)z') \geq \alpha f(z) +(1- \alpha)f(z')$$

let $\alpha z +(1- \alpha)z'=z''\in \mathbb{R^+}$ where by necessity $z\leq z'' \leq z'$

Since cost of $z=w$ is fixed at some $w \in \mathbb{R^+}$

we can rewrite our cost function as $C^f(q)$ which is convex if for $\alpha \in [0,1]$:

$$C(\alpha q +(1- \alpha)q') \leq \alpha c(q) +(1- \alpha)c(q')$$

Now, since $f$ is concave, it cannot be the case that $f$ experiences increasing returns to scale. Then $f$ has constant returns to scale or diminishing returns to scale.

Then:

Since $z''$ is clearly feasible then we have that:

$$C(w,q'') \leq w*z''$$ $$=\alpha w*z + (1- \alpha) w*z'$$ $$= \alpha c(w,q) + (1- \alpha) c(w,q')$$

Does this look correct?

Given the fixed input price $w$, the cost function can be written as $$C(q)=f^{-1}(q)\times w$$ where $f^{-1}$ is the inverse of the production function $f$. From the discussion here, one can conclude that the inverse of a concave strictly increasing function is convex. Thus, $C(q)$ is convex as well.
Going back to your approach, you might like to have this clearly stated. Let $q''=\alpha q + (1-\alpha)q'$, $f(z)=q$, $f(z')=q'$, and $f(z'')=q''$ Then \begin{align} f(z'')&=&q''\\ &=&(\alpha q + (1-\alpha)q')\\ &=&\alpha f(z)+(1-\alpha) f(z')\\ &\leq&f(\alpha z +(1-\alpha z')) \end{align} Then $f(z'')\leq f(\alpha z +(1-\alpha z'))$ implying that $\alpha z +(1-\alpha z')\geq z''$ since $f$ is (strictly) increasing. Hence, $$z''w=C(q'')\leq \alpha C(q)+ (1-\alpha) C(q')$$
• But isn't $f$ here only the left inverse of $C$? It does work as an inverse on the image, but not for arbitrary production plan that is not in the image? Since it describes the minimal costs for attaining given level of output (but there may be other, worse, plans that attain the same output). And then "the inverse of a concave strictly increasing function is convex" does not apply, does it? – kitsune Oct 14 '17 at 5:05