# Interior Solution for profit maximisation problem

A function $$c: \mathbb{R}^K_+ \xrightarrow{} \mathbb{R}_+$$ is is said to be a cost function if

1. The value of function $$c$$ at $$y = \textbf{0}$$ is $$0$$: $$c(\textbf{0}) = 0$$
2. $$c$$ is continuous on the domain $$\mathbb{R}^K_+$$, strictly increasing and strictly convex on $$\mathbb{R}^K_{+}$$
3. For any $$p \in \mathbb{R}^K_{++}$$, the set $$B(p) = \{y \in \mathbb{R}^K_+| c(y) \leq p \cdot y\}$$ is compact, the interior of $$B(p)$$ is nonempty.

Suppose a firm has cost function $$c(\cdot)$$ and $$p\in \mathbb{R}^K_{++}$$ is the market price. What sort of topological assumption I need to add to guarantee the existence of an interior solution to the profit maximisation problem? $$\pi (p) = \max_{y\in\mathbb{R}^K_+} \left\{ p\cdot y - c(y) \right\}$$

Remark 1: From assumption (2) and (3), the existence of solution is guaranteed. Due to strict convexity, uniqueness is also guaranteed. However, the solution is probably not inside the interior of $$\mathbb{R}^K_{+}$$. I could not find an example for this at the moment.

Remark 2: My prototype cost function to look at is $$c(y) = y_1^2+y_2^2$$. However, the cost is not necessarily to be additive separable.

• To clarify the question mathematically: by guarantee, do you mean a sufficient condition? Then specifying a particular cost function will work, but is not very general. More general might be something like: cost increases slower than linearly near 0, eventually faster than linearly. How do you define a topological assumption? What makes it topological? Mar 31, 2021 at 6:52
• @SanderHeinsalu Yes, by guarantee I mean sufficient condition. Topological assumption, from my viewpoint, may include assertions about openness, closedness and even convexity of certain sets but not differentiability assumption on the cost function. The prototype function is just an example to look at and learn about the minimal set of conditions that ensures interior solution. Mar 31, 2021 at 6:55
• Differentiability is not needed. Something about the growth rate of the cost probably is, vaguely along the lines: there exist k, K such that c(1)-c(0)<k<c(K+1)-c(K). One way to approach it is from the other end: find counterexamples of costs that satisfy your assumptions but that generate corner solutions. Your assumption 3. may already be sufficient. Mar 31, 2021 at 7:04
• @SanderHeinsalu I just learn that convex function always have right (and left) partial derivatives. Let $\partial^+_i c(y_i,y_{-i})$ be the right partial derivative in the $y_i$ direction. A sufficient condition could be $\partial^+_i c(0,y_{-i}) = 0$ for all $i$. However, as you suspected, (1) (2) and (3) are perhaps sufficient already and adding extra assumption just complicates the problem. I spend days finding a counter-example but I could not. Could you find one? Mar 31, 2021 at 8:48
• Could you include a source for your definition of cost function? The cost function of a firm is often (normally?) understood to be the function determining the minimum cost of producing any level of output given the costs of factor inputs. Such a function could be either concave or convex depending on whether there are economies or diseconomies of scale in production. Mar 31, 2021 at 11:21

Your assumption 3 is compatible with corner solutions of the kind $$y_i=0$$ for some $$i$$, and is not sufficient to avoid corner solutions for some prices small enough.
With production functions it is common to assume Inada's conditions to avoid corner solutions. With cost functions, such conditions are quite naturally expressed as $$\lim_{y_i\rightarrow 0} \frac{\partial c}{\partial y_i}(y_i,y_{-i})=0$$ $$\lim_{y_i\rightarrow +\infty} \frac{\partial c}{\partial y_i}(y_i,y_{-i})=+\infty.$$
In your discussion, you proposed the cost function $$c(y_1,y_2)=y_1^2+y_2^2+2y_1$$ to illustrate a corner solution. This function does not satisfy the first Inada conditions wrt $$y_1$$, hence the corner solution.