Home / Glossary / Cobb–Douglas Production Function
EconomicsCobb–Douglas Production Function
By Sitraka Forler · Lecturer, Durham Business SchoolUpdated 13 September 2026 About this site
Output as a function of capital and labour with constant returns.
Y = A·Kᵅ·Lᵝ models how capital (K) and labour (L) combine to produce output, with A as total factor productivity. When α + β = 1 there are constant returns to scale. The exponents are the income shares of each factor.
One surface, four contour lines: the Cobb-Douglas isoquant map
Each curve joins every capital and labour mix that produces one output level. Brighter blue means more output.
Y = K0.30 · L0.70, and the gold bundle K = 8, L = 2 makes Y = 3.03
doubling both inputs doubles output: 160.30 · 40.70 = 6.06 = 2 × 3.03: constant returns
The intuition
The Cobb-Douglas production function is a compact recipe for turning inputs into output: Y = A · K^α · L^β, where K is capital, L is labour and A is technology. The exponents are the heart of it. They are output elasticities: raise capital by 1% and output rises by about α%, holding labour fixed. When α + β = 1 the recipe has constant returns to scale, so doubling every input exactly doubles output. That single restriction makes the exponents measurable: in competitive markets, α and β are precisely the shares of income paid to capital and labour.
In the chart above, each blue curve is an isoquant: the full menu of capital and labour mixes that produce the same output, with brighter curves marking higher output levels. Sliding along one curve keeps production constant while machines replace workers, and the bend tells you how costly that swap becomes: the steeper the curve, the more labour you must give up for the next unit of capital. Drag the capital share α and the whole map pivots, because a capital-heavy technology squeezes more output from the same machines.
Economists reach for this form first because it is honest with data and generous with algebra. Take logarithms and it becomes a straight line, ln Y = ln A + α ln K + β ln L, so the exponents can be estimated by ordinary regression. Better still, the share of income going to labour hovered near two thirds for much of the twentieth century, exactly what a Cobb-Douglas economy with β close to 0.7 predicts. Few models earn their popularity so directly.
Formula / theory
Y = A · Kᵅ · Lᵝ (constant returns when α + β = 1)
In Python
Y = A * K**alpha * L**beta # log-linear form for regression import numpy as np ln_Y = np.log(A) + alpha * np.log(K) + beta * np.log(L)
From inputs to output, and back to income shares
- Set the technology: A = 1 and capital share α = 0.3, so the labour exponent is β = 1 − 0.3 = 0.7 and the function is Y = K^0.3 · L^0.7.
- Choose round inputs: K = 8 units of capital and L = 27 workers. Capital term: 8^0.3 = 1.866.
- Labour term: 27^0.7 = 10.045.
- Multiply the two terms: Y = 1.866 × 10.045 ≈ 18.74 units of output.
- Double both inputs to K = 16 and L = 54: (2K)^0.3 · (2L)^0.7 = 2^0.3 · 2^0.7 · K^0.3 · L^0.7 = 2 × 18.74 = 37.48. Output exactly doubles: constant returns to scale.
- Split the income: in competitive markets labour earns β · Y = 0.7 × 18.74 ≈ 13.12 and capital earns α · Y = 0.3 × 18.74 ≈ 5.62. The two payments sum to 13.12 + 5.62 = 18.74, exhausting output exactly.
With constant returns, the exponents are not abstract curvature settings: they are the exact fractions of output paid to each input, which is why α near 0.3 matches capital's observed share of national income.
Common pitfalls
- Reading α as the amount or share of capital used in production. It is an output elasticity: the percentage rise in output caused by a 1% rise in capital, with labour held fixed.
- Assuming constant returns automatically. Nothing forces α + β = 1: if the exponents sum to more than one, doubling inputs more than doubles output, and paying each factor its marginal product would more than exhaust total output.
- Forgetting that Cobb-Douglas fixes the elasticity of substitution at exactly 1. Where inputs are genuinely hard to substitute, such as energy in the short run, a CES function is the safer choice.
- Estimating the exponents by regressing log output on log inputs while ignoring simultaneity: firms choose their inputs after observing productivity shocks, so ordinary least squares estimates of α and β are biased.
Frequently asked questions
What does the Cobb-Douglas production function show?
It shows how much output an economy or firm can produce from given amounts of capital and labour, using the formula Y = A · K^α · L^β. The exponents α and β measure how strongly output responds to each input, and A captures technology or productivity. It is the standard baseline model of production in economics, used in growth theory, macroeconomic models and empirical work.
What do alpha and beta mean in the Cobb-Douglas function?
Alpha and beta are output elasticities: a 1% increase in capital raises output by about α%, and a 1% increase in labour raises it by about β%. Under constant returns and competitive markets they also equal the income shares of each input, which is why α is typically set near 0.3 and β near 0.7, matching the split between capital and labour income in national accounts.
What does constant returns to scale mean in Cobb-Douglas?
Constant returns to scale means that scaling every input by the same factor scales output by exactly that factor, and in Cobb-Douglas it holds precisely when α + β = 1. Doubling both inputs gives (2K)^α · (2L)^β = 2^(α+β) · K^α · L^β, which is twice the original output when the exponents sum to one. If they sum to more, returns are increasing; to less, decreasing.
Why is the Cobb-Douglas production function used so much?
Because it is tractable, estimable and empirically reasonable. Taking logarithms makes it linear, so α and β can be estimated by simple regression. Its predictions match a famous stylised fact: the income shares of labour and capital stayed roughly stable for much of the twentieth century. And its algebra stays clean inside larger models, from Solow growth to business cycle models, which keeps it the default first choice.
Test yourself
Further reading
Where to go deeper. Free means a full, legal copy is online.
The 1928 paper that fit output to capital and labour and gave the function its name.