Cost Functions

Cost functions are mathematical descriptions of how a firm's costs depend on the quantity of output it produces. At its most basic, a cost function answers: if the firm produces $q$ units, what is the total cost $TC(q)$? Splitting total cost into components yields intuition and decision rules that managers and economists use.

Key pieces and why they matter

• •Total Cost ($TC(q)$): The total economic cost of producing $q$ units, usually measured in dollars. Example: if producing nothing still requires some expenses (like rent), then $TC(0)>0$. Concretely, consider a simple quadratic cost function

Numerically, $TC(0)=100$ (the firm has a $100 fixed cost if it shuts down output), and$TC(10)=100+5(10)+0.5(10)^2=100+50+50=200$.

• •Fixed Cost ($FC$): Costs that do not vary with $q$ in the short run (e.g., rent, machinery that must be paid for whether you use it or not). In the example, $FC=100$. Fixed cost is the vertical intercept of $TC(q)$.

• •Variable Cost ($VC(q)$): Costs that vary with output (e.g., raw materials, hourly labor). By definition $VC(q)=TC(q)-FC$. In the example,

For $q=10$, $VC(10)=5(10)+0.5(100)=50+50=100$.

• •Average Cost measures: These let you compare per-unit costs across output levels.
• •Average Total Cost (ATC or $AC$): $AC(q)=\dfrac{TC(q)}{q}$. For the example,

If $q=10$, $AC(10)=\frac{100}{10}+5+0.5(10)=10+5+5=20$.

• •Average Fixed Cost ($AFC$): $AFC(q)=\dfrac{FC}{q}=\dfrac{100}{q}$. For $q=10$, $AFC(10)=10$.

• •Average Variable Cost ($AVC$): $AVC(q)=\dfrac{VC(q)}{q}=5+0.5q$. For $q=10$, $AVC(10)=5+5=10$.

Notice $AC(q)=AFC(q)+AVC(q)$ numerically: $20=10+10$.

• •Marginal Cost ($MC$): The instantaneous additional cost of producing one more unit. In calculus terms (see Derivatives (d2)), marginal cost is the derivative of total cost with respect to $q$:

For our example,

Numerically, $MC(10)=5+10=15$. Marginal cost approximates the cost of the 11th unit; using discrete difference, $TC(11)-TC(10)=\big(100+5(11)+0.5(11)^2\big)-200=100+55+60.5-200=15.5$, close to $MC(10)=15$ since the function is smooth.

Short-run vs Long-run

• •Short-run: At least one input is fixed (hence fixed cost exists). The short-run cost functions (FC, VC, MC, AC) are conditional on the current plant size or capacity.

• •Long-run: All inputs are variable. There is no fixed cost in the long run for the firm's planning problem. Long-run cost (LRTC or $LAC$) is the envelope of short-run average costs across all possible plant sizes. Long-run cost determines optimal scale, which leads to the concept of economies of scale.

Economies of scale

• •If $LAC(q)$ decreases as $q$ increases over some range, the firm enjoys economies of scale (average cost per unit falls with output). If $LAC(q)$ increases with $q$, there are diseconomies of scale. If $LAC(q)$ is flat, constant returns to scale.

Why this matters: Marginal cost interacts with price and marginal revenue to determine optimal output (see Optimization Introduction (d3)). Average cost tells you whether a firm covers its costs at a given price and whether expanding output reduces unit costs.

Marginal cost (MC) is central because in competitive markets the profit-maximizing condition often equates price to marginal cost (when firms are price takers). Average costs determine profitability and long-run entry/exit.

Formal relationships and calculus intuition (reference: Derivatives (d2))

• •Marginal cost as a derivative: Given $TC(q)$ differentiable,

Example: If $TC(q)=50+2q+0.1q^2$, then

Numerically, $MC(20)=2+0.2(20)=2+4=6$. So the additional cost at output 20 is $6 per extra unit.

• •Marginal cost vs average cost slope: A basic calculus result (related to the derivative of a quotient) tells us that $MC(q)$ intersects $AC(q)$ at the $AC$ minimum. Consider $AC(q)=TC(q)/q$. Differentiate using the quotient rule (or product rule if you write $AC(q)=TC(q)\cdot q^{-1}$):

Setting $AC'(q)=0$ gives $q\cdot MC(q)-TC(q)=0\Rightarrow MC(q)=TC(q)/q=AC(q)$. Thus the stationary point of $AC$ occurs where $MC=AC$. If that point is a minimum, MC crosses AC from below.

Concrete numeric check: Take $TC(q)=100+5q+0.5q^2$ as before. Then $AC(q)=100/q+5+0.5q$ and $MC(q)=5+q$. Solve $MC(q)=AC(q)$:

Numerically check: $MC(14.142)\approx5+14.142=19.142$. And $AC(14.142)=100/14.142+5+0.5(14.142)\approx7.071+5+7.071=19.142$, matching.

• •Behavior of MC and AC curves: Because $AFC$ falls with $q$ (since $AFC=FC/q$), $AC$ initially falls if $AVC$ is relatively flat and $AFC$ dominates. But if $MC$ is increasing (e.g., due to diminishing marginal returns), $MC$ eventually pulls $AC$ up. Graphically, $MC$ typically has a U-shape and crosses the U-shaped $AC$ curve at its minimum.

• •Marginal cost from variable cost: Since $FC$ is constant, $MC(q)=TC'(q)=VC'(q)$. So for calculus problems it's often enough to specify $VC(q)$. Example: If $VC(q)=20q+0.4q^2$, then $MC(q)=20+0.8q$. For $q=25$, $MC(25)=20+0.8(25)=20+20=40$.

Local decision rules and short-run shutdown condition (link to Optimization Introduction (d3))

• •Profit or loss short-run: Profit is $\pi(q)=TR(q)-TC(q)$ where $TR(q)=p\cdot q$ in perfect competition. First-order condition (from Optimization Introduction (d3) and derivatives) for an interior optimum: $\pi'(q)=0\Rightarrow p-TC'(q)=0\Rightarrow p=MC(q)$.

Example numeric decision: If price $p=19$ and $MC(q)=5+q$, then optimal output solves $5+q=19\Rightarrow q=14$. This matches the earlier $AC$ minimum example and shows how calculus directly produces the optimal output.

• •Shutdown rule: In the short run, if price $p$ is below the minimum of $AVC(q)$, the firm should shut down and produce $q=0$ (it would minimize losses by paying fixed costs only). If $p$ exceeds minimum $AVC$, produce where $p=MC$.

Concrete numbers: For $VC(q)=5q+0.5q^2$, $AVC(q)=5+0.5q$ has minimum at $q=0$ (since it's increasing here) with $AVC(0)=5$. If market price $p=4$, since $p<5$ the firm should shut down. If $p=10$, solve $MC=5+q=10\Rightarrow q=5$.

This section shows how derivatives (Derivatives (d2)) and optimization first-order conditions (Optimization Introduction (d3)) translate to managerial rules: set output where $p=MC$, check $AVC$ to decide shutdown, and use $AC$ to check profitability.

Understanding the distinction between short-run and long-run cost structures is essential for planning capacity and for understanding industry dynamics (entry, exit, and firm size).

Short-run cost structure

• •In the short run, at least one input is fixed (e.g., capital). The firm faces a specific $TC_{SR}(q)=FC+VC_{SR}(q)$ for its current plant. Fixed cost $FC$ does not change with $q$ in the short run. Consequently, $AFC(q)=FC/q$ declines as $q$ rises.

Example: Suppose a firm has a factory requiring $FC=200$ and variable cost $VC_{SR}(q)=10q+0.2q^2$. Then

Numerical check: at $q=20$, $TC_{SR}(20)=200+200+80=480$ (wait: compute carefully: $10*20=200$, $0.2*400=80$, so $TC_{SR}=480$), $MC_{SR}(20)=10+0.4(20)=18$, $AC_{SR}(20)=200/20+10+0.2(20)=10+10+4=24$.

• •The short-run average cost curve for a given plant is typically U-shaped because $AFC$ falls with $q$ (pulling $AC$ down at low q) while rising marginal costs cause $AVC$ to rise and eventually pull $AC$ up.

Long-run cost structure and the envelope theorem

• •Long-run costs allow adjustment of all inputs. For each target output $q$, the firm can choose the plant size or input mix that minimizes cost. Long-run total cost $LTC(q)$ equals the minimal short-run total cost across all feasible plant sizes. Graphically, the long-run average cost curve $LAC(q)$ is the lower envelope of the short-run average cost curves for different plant sizes.

Algebraically, suppose plant size is parameterized by $k$ (e.g., capacity). For each $k$ we have $TC(q;k)=FC(k)+VC(q;k)$. Then

Example (simplified): Suppose there are two discrete plant sizes:

• •Small plant: $TC_S(q)=100+6q+0.2q^2$ (good for low q)
• •Large plant: $TC_L(q)=400+3q+0.1q^2$ (more fixed cost but lower variable cost)

For $q=10$, $TC_S(10)=100+60+20=180$, $TC_L(10)=400+30+10=440$. So the small plant is cheaper at low output. For $q=200$, $TC_S(200)=100+1200+8000=9320$, $TC_L(200)=400+600+4000=5000$; the large plant is cheaper at high output. The long-run cost at each $q$ picks the cheaper plant, creating an $LAC(q)$ that can fall then rise depending on technology.

Economies of scale (technical/market meanings)

• •Economies of scale: If $LAC(q)$ decreases as $q$ increases over some range, doubling output less than doubles cost (average cost falls). For example, if $LTC(q)=aq^\alpha$ with $\alpha<1$, then $LAC(q)=aq^{\alpha-1}$ falls with $q$.

Concrete numeric instance: Take $LTC(q)=10q^{0.8}$. Then $LAC(q)=10q^{-0.2}$. For $q=100$, $LAC(100)=10(100)^{-0.2}=10/(100^{0.2})=10/(\approx2.5119)\approx3.98$. For $q=1000$, $LAC(1000)=10(1000)^{-0.2}=10/(1000^{0.2})=10/(\approx3.981)\approx2.51$. So average cost falls with scale: economies of scale.

• •Sources of economies of scale: specialization, indivisibilities in capital (a huge machine that spreads cost across many units), bulk buying of inputs, or spreading fixed costs. Sources of diseconomies: managerial coordination costs, congestion, or increasing input prices when scaling.

Returns to scale vs economies of scale distinction

• •Returns to scale (a production function concept) measures how output responds when all inputs scale. If doubling inputs more than doubles output, there are increasing returns to scale. This translates into cost language: increasing returns to scale in production typically imply economies of scale (falling $LAC$), but care is needed because input prices and discrete plant options can alter the mapping.

Long-run planning decisions

• •Firms choose the plant size to minimize average cost for their expected output in the long run. If demand grows, a firm may invest in a larger plant to move to a lower $LAC$ region.

• •Industry implications: If $LAC$ falls over a wide range such that large firms have much lower $LAC$ than small ones, the industry is natural monopoly-prone (one large firm can supply at lower average cost). If $LAC$ is fairly flat, many firms can coexist.

Cost functions are used across microeconomics, managerial economics, and public policy. Here are concrete applications and how this connects to downstream topics.

Pricing and output decisions in market structures

• •Perfect competition: Firms are price takers. Profit maximization uses calculus (Optimization Introduction (d3)): set $p=MC(q)$ if $p\geq AVC_{min}$; otherwise shut down. Example: If market price $p=30$ and $MC(q)=10+0.5q$, solve $10+0.5q=30\Rightarrow q=40$. If $AVC_{min}=12$ and $p=11$, firm should shut down.

• •Monopoly/monopolistic competition: Marginal cost still matters, but the profit-maximizing rule uses marginal revenue $MR(q)$: set $MR(q)=MC(q)$. For a linear demand $P(q)=100-2q$, $TR(q)=100q-2q^2$ so $MR(q)=100-4q$. With $MC(q)=20+q$, the optimum solves $100-4q=20+q\Rightarrow5q=80\Rightarrow q=16$, charging $P(16)=100-32=68$.

Cost-benefit, entry/exit, and long-run industry supply

• •Entry and exit: In the long run, firms enter if $P>LAC(q^*)$ for potential entrants and exit if $P<LAC(q^*)$ for incumbents. For example, if the competitive market price converges to the minimum of $LAC$, firms earn zero economic profit in a free-entry long-run equilibrium.

• •Natural monopoly and regulation: If $LAC$ declines across the entire relevant output range (due to large fixed costs), a single firm produces at lower cost than many small ones. Regulators may set price equal to $AC$ to allow zero economic profit, or to $MC$, sometimes with subsidies if $MC<AC$ for all q.

Cost estimation and managerial accounting

• •Empirical cost estimation fits functional forms (linear, quadratic, translog) to observed costs. The marginal cost derived from estimated $TC(q)$ then informs pricing decisions. Example: if regression yields $TC(q)=150+8q+0.3q^2$, managers compute $MC(q)=8+0.6q$ to set incremental pricing.

Network industries and scale economies

• •In software or platform firms, marginal cost of an additional customer can be near zero (digital goods). Here $MC\approx0$ and average cost falls dramatically with output (huge economies of scale). Pricing strategies (freemium, subscription) exploit low marginal cost and high fixed development cost.

Connection to welfare analysis

• •In public policy, comparing $P$ to $MC$ is central for efficiency: price equals marginal cost is allocatively efficient. When $MC$ is below $AC$ due to high fixed costs, setting price at $MC$ can require subsidy; regulators weigh efficiency vs distribution.

Downstream topics that use cost function mechanics

• •Production theory: returns to scale and cost functions are duals; cost minimization problems produce cost functions and conditional factor demands.

• •Firm dynamics and industrial organization: long-run cost shapes market structure (number and scale of firms). Empirical IO models estimate $LAC$ to understand barriers to entry.

• •Game theory and strategic pricing: knowledge of rival cost curves informs price competition (Bertrand) and quantity competition (Cournot).

Concrete managerial checklist

• •Compute $MC(q)=TC'(q)$ and $AC(q)=TC(q)/q$ for candidate $q$.
• •If price is given, check shutdown: if $p<\min AVC$, produce $q=0$; otherwise choose $q$ with $p=MC(q)$.
• •For long-run planning, compute $LAC(q)$ by minimizing $TC(q;k)$ over plant choices; identify range of $q$ where $LAC$ falls (economies of scale).

Every formula in this lesson ties back to calculus (Derivatives (d2)) for rates of change and to optimization first-order conditions (Optimization Introduction (d3)) for choosing $q$ to maximize profit or minimize cost. These tools give actionable rules for pricing, capacity choice, and policy design.