Producer Theory and Cost Functions Explained: How Firms Choose Inputs and What a Wage Rise Costs Them

Understanding producer theory and cost functions — simply put

Producer theory describes how a firm decides how much to produce and which inputs to use. A firm can make the same output with different combinations of labor, machines, energy and materials. Given the prices of those inputs, one combination is cheapest. The cost function records that cheapest cost for every output level and every set of input prices. Once you know it, you can say how a firm's costs respond to a wage rise, an energy shock or a tax, and how much of each input it will want.

How does it actually help?

Take a firm whose output depends on labor and capital through a Cobb-Douglas production function:

The firm pays a wage w per hour of labor and a rental rate r per hour of capital. Its problem is to produce a target output q at the lowest possible total cost wL + rK. The solution is the cost function C(w, r, q). When α + β = 1, so that doubling both inputs doubles output (constant returns to scale), cost is proportional to output, and the cost of one unit is:

That last point is the useful one. A firm whose wage bill is half its costs has α = 0.5, and a 10% wage rise raises its unit cost by roughly 0.5 × 10% = 5%, not 10%, because labor is only part of the cost and the firm partly substitutes toward capital.

Why does this matter?

Cost functions are how economists and regulators turn a change in input prices into a change in prices, employment and profits. Competition authorities use them to judge whether mergers create cost savings. Energy regulators use them to set allowed revenues. Minimum wage commissions use them to estimate how much a wage floor raises costs in exposed sectors. Productivity analysis, which asks why output per hour grows faster in some countries than others, rests on the same production functions.

What makes the cost function reliable?

The firm has to be minimizing cost, taking input prices as given, and the technology has to be well behaved: more inputs never reduce output, and inputs can be substituted for one another smoothly. Under those conditions the cost function has properties that can be checked against data, which is what makes it useful in practice. The full list, and what fails without each condition, follows the example.

The takeaway

The production function describes what is technically possible; the cost function describes what is economically sensible. Together with Shephard's lemma they turn input prices into input demands and unit costs. The worked example applies them to a real wage shock, and the formula section derives every result.

A real example: Germany's €12 minimum wage

Germany's statutory minimum wage was €9.82 an hour in January 2022. It rose to €10.45 on 1 July 2022 and, under legislation approved by the Federal Cabinet, to €12 on 1 October 2022, an increase the government said would benefit more than six million workers. The Minimum Wage Commission later set further increases to €12.41 from January 2024 and €12.82 from January 2025.

Consider a firm whose workers are paid at the minimum wage and whose labor costs are half its total costs (α = 0.5, β = 0.5). This cost share is an illustrative assumption, typical of labor-intensive services such as cleaning, catering and hospitality; the actual share varies by firm and sector. Capital costs are held fixed. With constant returns to scale, two results follow from the cost function.

Reading the result

Unit cost. With a Cobb-Douglas cost function, unit cost scales with w^α. A 22.2% wage rise therefore raises unit cost by 1.222^0.5 − 1 ≈ 10.5%. The October 2022 step alone, from €10.45 to €12, was a 14.8% wage rise and a 7.2% cost rise.

Labor per unit of output. By Shephard's lemma the firm's labor requirement per unit of output is the derivative of unit cost with respect to the wage. For Cobb-Douglas it scales with w^(α−1), so a 22.2% wage rise cuts labor per unit by 1 − 1.222^(−0.5) ≈ 9.5%. This is conditional labor demand: it holds output fixed and captures only the switch toward capital, such as self-service machines or fewer staff per shift.

What the table does not say. It does not predict that total employment falls by 9.5%. Total employment also depends on what happens to output, which depends on how much of the cost rise is passed on to prices and how much customers cut back. If the firm sets wages with some market power, which is common in low-wage labor markets, a minimum wage can even raise employment. Research on Germany's original 2015 minimum wage found little effect on total employment and a shift of workers toward more productive firms, and the classic difference-in-differences study of New Jersey's 1992 increase reached a similar conclusion. The cost function measures the pressure on costs; how that pressure plays out in employment is an empirical question.

The assumptions

Cost minimization. The firm uses the cheapest input mix for whatever output it produces. This holds for any firm maximizing profit, whatever its market power in the product market, because producing a given output at more than minimum cost leaves profit on the table. If managers pursue other goals, observed costs lie above the cost function and the derived input demands are wrong.

Price-taking in input markets. The firm treats w and r as given. If the firm's own hiring moves the wage it pays (monopsony), the relevant cost of an extra hour is above w, and the standard input demand formulas understate the cost of labor and mispredict the response to a wage floor.

Monotonicity. More of any input never reduces output. Without it, a cost-minimizing firm would never use some input levels and the cost function would not be increasing in output.

Quasi-concavity of the production function (convex isoquants). Balanced input mixes are at least as productive as extremes. This makes the cost-minimizing input mix unique and makes input demands slope downward in their own price. Without it, small price changes can cause jumps from one input mix to another.

Returns to scale. The example assumes constant returns, α + β = 1, so unit cost does not depend on output. With increasing returns (α + β > 1) unit cost falls as output grows; with decreasing returns it rises. The scale assumption affects the cost-output relationship, not the input substitution results.

The formula

The derivations below are optional for readers who only need the intuition.

The firm's cost minimization problem is:

The Lagrangian is ℒ = wL + rK − λ[f(L, K) − q], where λ is the Lagrange multiplier. The first-order conditions are w = λ f_L and r = λ f_K, where f_L and f_K are the marginal products of labor and capital. Dividing one by the other gives the tangency condition:

The multiplier λ equals marginal cost, ∂C/∂q: it measures how much the minimum cost rises when the output target rises by one unit.

Cobb-Douglas solution. With f(L, K) = A L^α K^β, the tangency condition gives K/L = (β/α)(w/r). Substituting into the output constraint and solving:

Shephard's lemma. By the envelope theorem, the derivative of the cost function with respect to an input price equals the conditional demand for that input:

Multiplying by w/C converts Shephard's lemma into a statement about shares: the elasticity of cost with respect to the wage equals labor's share in cost, wL*/C. For Cobb-Douglas with constant returns that share is exactly α, which is why the worked example could read the cost response directly off the labor share. The own-wage elasticity of conditional labor demand is −(1 − α), here −0.5.

Properties of every cost function. Whatever the technology, a cost function derived from cost minimization is nondecreasing in input prices, homogeneous of degree one in input prices (doubling all input prices doubles cost), and concave in input prices. Concavity is the producer-side counterpart of the negative substitution effect in consumer theory: conditional input demands never rise with their own price. These restrictions can be tested with regression on firm or industry data, and flexible forms such as the translog cost function are estimated precisely so they can be.

Profit maximization and Hotelling's lemma. A price-taking firm facing output price p chooses q to maximize p·q − C(w, r, q), giving p = ∂C/∂q: price equals marginal cost. The resulting profit function π(p, w, r) satisfies Hotelling's lemma:

The difference between conditional demand L*(w, r, q) and unconditional demand L(p, w, r) is the scale effect noted in the example: a wage rise raises marginal cost, the firm produces less, and it uses less labor on top of the substitution toward capital. The same logic underlies the consumer theory duality between Hicksian and Marshallian demand.

Short run and long run. If capital is fixed at K̄ in the short run, the firm can adjust only labor, and short-run cost C_S(w, r, q; K̄) lies on or above long-run cost, touching it at the output level for which K̄ is optimal. The long-run average cost curve is the lower envelope of the short-run curves.

When this isn't the right tool

Cobb-Douglas fixes the elasticity of substitution at one. Labor and capital may be much harder to substitute in the short run. A CES or translog cost function estimated on real cost data gives a more credible response; with less substitution, the unit-cost rise is closer to the full labor share times the wage rise.

Labor markets are not always competitive. Where employers have wage-setting power, a minimum wage changes the marginal cost of labor in a way the price-taking model does not capture.

Cost is not the whole story. The model takes output demand as given. To trace a cost shock through to prices, quantities and other industries, the cost function has to be combined with a model of the product market, such as supply and demand or a general equilibrium model.

Further reading

• Consumer Theory and Duality Explained — internal
• General Equilibrium and the Welfare Theorems Explained — internal
• Difference-in-Differences Explained — internal
• Supply and Demand Explained — internal
• OLS Regression Explained — internal
• German Federal Government, the twelve-euro minimum wage — external reference
• Minimum Wage Commission — external reference

Minimum wage levels are from the German Federal Government and the Minimum Wage Commission. The 50% labor cost share is an illustrative assumption, and the cost and labor effects are EconoLens calculations from a Cobb-Douglas cost function, not measured outcomes.