Economic growth · Lesson 2 of 14

The production function: returns to scale, Cobb–Douglas and technical change

Q: What is a production function?

A production function says how much output an economy or firm can produce from given amounts of inputs, usually capital K and labour L, and a level of technology A. In growth theory it is written Y = F(K, L, A), and the standard example is the Cobb–Douglas function Y = A K^α L^(1−α).

Q: What are constant returns to scale, and why does growth theory assume them?

Constant returns to scale means that multiplying all inputs by the same factor multiplies output by that factor: double K and L and Y doubles. Growth theory assumes it because an economy can be replicated: two identical economies produce twice as much. It also lets output per worker depend only on capital per worker, y = f(k), which is what makes the Solow model solvable.

Q: What is the Cobb–Douglas production function?

Y = A K^α L^β. The exponents are the output elasticities of capital and labour. If α + β = 1 there are constant returns to scale, and if inputs are paid their marginal products, capital receives a share α of income and labour a share 1 − α. Cobb and Douglas fitted it to US manufacturing data in 1928.

Q: What are the Inada conditions?

They are limits on the shape of the production function that guarantee a steady state exists: output is zero with no capital, the marginal product of capital is positive and falling, it is infinite as capital approaches zero and goes to zero as capital becomes very large. Cobb–Douglas with capital's exponent between 0 and 1 satisfies them.

Q: What is the difference between Hicks-neutral, Harrod-neutral and Solow-neutral technical change?

Hicks-neutral change multiplies the whole function, Y = A F(K, L), and leaves the ratio of marginal products unchanged at a given capital–labour ratio. Harrod-neutral (labour-augmenting) change makes labour more effective, Y = F(K, AL), and keeps the capital–output ratio constant at a given interest rate. Solow-neutral (capital-augmenting) change makes capital more effective, Y = F(AK, L). For a Cobb–Douglas function all three are the same thing written differently.

Q: Why must technical progress be labour-augmenting for steady growth?

Uzawa (1961) showed that a neoclassical model can have a balanced growth path, with all variables growing at constant rates, only if technical progress is labour-augmenting (Harrod-neutral), apart from the Cobb–Douglas case where the forms coincide. That is why the Solow model with technical progress writes output as F(K, AL).

Every growth model begins with the same sentence: output depends on capital, labour and know-how. The shape of that dependence, how output responds when you add more of one input or all of them, decides what the model can say. This lesson builds the shape from scratch and introduces the function almost every exam question uses.

← Economic growth moduleLesson 2 of 14 · about 45 minutes

The short answer

What is a production function, and why does growth theory assume constant returns?

A production function states how much output Y an economy can produce from capital K, labour L and a level of technology A. Growth theory almost always uses the Cobb–Douglas form because it has constant returns to scale, diminishing returns to each input, and exponents that read directly as factor shares.

Y = F (K, L, A) Y = A K^{α} L^{1 - α} y = \frac{Y}{L} = A k^{α}, k = \frac{K}{L}

Returns to scale ask what happens to output when all inputs are multiplied by λ. Output is multiplied by λ^a+b in Cobb–Douglas, so constant returns means the exponents sum to 1.
Diminishing returns ask what happens when one input is added with the others fixed: its marginal product falls. Both can hold at once.
Constant returns give the intensive form: output per worker depends only on capital per worker, y = f(k). The Solow model rests on this step.
Cobb–Douglas exponents are factor shares: capital earns α of income and labour 1 − α, when inputs are paid their marginal products.

A question to hold on to

In the Solow growth model, the assumption of constant returns to scale means that: (A) the number of workers in an economy does not affect the relationship between output per worker and capital per worker; (B) all economies have the same amount of capital per worker; (C) the steady-state level of output is constant regardless of the number of workers; (D) the saving rate equals the constant rate of depreciation.

CUET-PG 2020Solved at the end ↓

New to this? The few things you need first

Marginal product: The extra output from one more unit of an input, the derivative of output with respect to that input, others held fixed.
Exponents: x^ax^b = x^a+b, and (λx)^a = λ^ax^a. These two rules do almost all the work in this lesson.
Growth and levels: Lesson 1: growth is the percentage change in a level; per-head growth is output growth minus population growth.

The idea

Output from inputs, and diminishing returns

Write output as $Y = F (K, L)$ : more capital or more labour gives more output. Two properties are almost always assumed.

Positive marginal products. $\partial Y / \partial K > 0$ and $\partial Y / \partial L > 0$ . A bit more of either input never reduces output.
Diminishing marginal products. $\partial^{2} Y / \partial K^{2} < 0$ . Each extra machine adds less than the last, because the workforce that operates it is fixed. A thousand machines for one worker are useless.

Diminishing returns to one input is a statement about adding it while holding the other fixed. Do not confuse it with returns to scale, which scales every input together. A function can have both: diminishing returns to each input and constant returns to scale.

Hyderabad 2019 · standard

A production function is y = −x³ + 6x² + 15x, where y is output and x is capital. For what value of x does the marginal product of capital start decreasing?

The marginal product.
$M P = \frac{d y}{d x} = - 3 x^{2} + 12 x + 15$
When is it decreasing?
When its own derivative is negative: $\frac{d ( M P )}{d x} = - 6 x + 12 < 0$ , so x > 2.
The turning point.
At x = 2 the marginal product is at its peak (it is 27 there) and it falls for every larger x.

Answer

x = 2.0. The marginal product rises up to 2 and declines after it, the “point of diminishing returns”. Hyderabad 2020 asks the same with y = −x³ + 18x² + 45x, which gives x = 6.

Interactive

Returns to scale

Now multiply both inputs by λ. If output is multiplied by λ, there are constant returns to scale (CRS). If by more, increasing; by less, decreasing. Formally, a function is homogeneous of degree h if

Homogeneity and returns to scale

F (λ K, λ L) = λ^{h} F (K, L)

h = 1: constant returns. h > 1: increasing. h < 1: decreasing. A degree-1 function also satisfies Euler's theorem: K·F_K + L·F_L = F.

For Cobb–Douglas $Y = A K^{a} L^{b}$ , scaling gives $A (λ K)^{a} (λ L)^{b} = λ^{a + b} Y$ , so the degree is a + b. Try the lab: set the exponents free and see what happens when both inputs double.

Constant returns: double both K and L and output exactly doubles. Pay each input its marginal product and capital receives 40% of output, labour 60%, the two adding to all of it.

Pause and predict

For Q = LK/(L + K), what is the degree of homogeneity?

Euler’s theorem, and why constant returns are assumed

If F is homogeneous of degree 1, then $K F_{K} + L F_{L} = F$ . Divide by Y and it says that capital’s share $F_{K} K / Y$ plus labour’s share $F_{L} L / Y$ adds to 1: if each input is paid its marginal product, payments exactly exhaust output, with nothing left over as profit. That is one reason constant returns are the natural assumption in a competitive economy. The other is replication: build a second identical economy and you can produce twice as much, so output cannot more than double, and the first economy should not be able to do worse.

UGC NET 2024 · hard

For Q = AK^αL^β with A, α, β > 0: (A) the degree of homogeneity is 1; (B) the output elasticity of capital is α; (C) it has constant returns to scale; (D) the marginal product of a factor equals its average product. Which are correct?

Degree.
It is α + β, which is 1 only if the exponents happen to sum to 1. So (A) and (C) are not true in general.
Elasticity.
Output elasticity of capital is (∂Q/∂K)(K/Q) = αAK^α−1L^β · K ÷ Q = α. (B) is true.
MP against AP.
MP_K = α·Q/K = α·AP_K. They are equal only if α = 1. So (D) is false in general.

Answer

Only (B). A statement that holds for every α and β is the elasticity one.

The workhorse

Cobb–Douglas and factor shares

The Cobb–Douglas function with constant returns is

Cobb–Douglas with constant returns

Y = A K^{α} L^{1 - α}

A is total factor productivity. α is capital's exponent, between 0 and 1.

It is the standard because it has four convenient properties at once:

Constant returns, since the exponents sum to 1.
Diminishing marginal products: $M P K = α Y / K$ falls as K rises.
Constant factor shares. Pay capital its marginal product and it receives $M P K \cdot K = α Y$ , a share α of output, whatever K and L are. Labour gets 1 − α. This is Kaldor’s fifth fact (Lesson 1) built into the function.
Growth rates add. Taking logs, $ln Y = ln A + α ln K + (1 - α) ln L$ , so $g_{Y} = g_{A} + α g_{K} + (1 - α) g_{L}$ . That equation is the whole of growth accounting (Lesson 3).

In rich economies capital’s share is usually put at about one-third, so α ≈ 1/3 is the conventional calibration. Cobb and Douglas themselves, fitting US manufacturing for 1899 to 1922, found labour’s exponent near three-quarters.

The key step

The intensive form: output per worker

With constant returns you may scale by any factor you like, so scale by 1/L. That turns the function into one in output per worker and capital per worker:

The intensive form

\frac{Y}{L} = F (\frac{K}{L}, 1) ⟹ y = f (k)

y = Y/L, k = K/L, and f(k) = F(k, 1). For Cobb–Douglas, f(k) = A k^α.

This is the step the opening question is about: under constant returns the size of the workforce does not matter, only how much capital each worker has. A country with twice the workers and twice the capital has the same output per worker. It is also what lets the Solow model be drawn on a two-dimensional diagram.

What shape should f(k) have?

f(0) = 0: nothing to work with, nothing produced.
f′(k) > 0 and f″(k) < 0: more capital per worker raises output per worker, with diminishing returns.
The Inada conditions: f′(k) → ∞ as k → 0, and f′(k) → 0 as k → ∞. The first says the first machines are extremely productive and the second that capital eventually adds almost nothing. Together they guarantee that the Solow model has a steady state (Lesson 5).

Cobb–Douglas with 0 < α < 1 satisfies all of these. If α = 1, so that f(k) = Ak, the second Inada condition fails: capital never has diminishing returns, and the model becomes the AK model of Lesson 10.

Pause and predict

With y = k^0.5 and 400 units of capital shared among 100 workers, output per worker is

Technology

Technical change and neutrality

Technology A has been a single number so far. But technical progress can enter the function in different ways, and which way decides whether the economy can grow steadily. There are three standard cases:

Kind	Form	What it does	What stays constant
Hicks-neutral	$Y = A F (K, L)$	Scales up all output; every input becomes equally more productive	The ratio of marginal products, at a given K/L
Harrod-neutral (labour-augmenting)	$Y = F (K, A L)$	Each worker becomes more effective, as if there were more workers	The capital–output ratio, at a given rate of return
Solow-neutral (capital-augmenting)	$Y = F (A K, L)$	Each machine becomes more effective, as if there were more capital	The labour–output ratio, at a given wage

For Cobb–Douglas the three are the same thing in different clothes: $A_{H} = A_{L}^{1 - α} = A_{K}^{α}$ , so you can move A from outside to inside the brackets by changing the exponent. With any other function they differ. That matters because of a famous result:

Uzawa's result

A neoclassical model has a balanced growth path, with output, capital and consumption all growing at constant rates, only if technical progress is labour-augmenting (Harrod-neutral), unless the function is Cobb–Douglas. That is why the Solow model with technology writes Y = F(K, AL), and why a CUET-PG question (2023) asks which neutrality the simple Solow model assumes: Harrod-neutral.

Solving the opening question

The opening question, solved

CUET-PG 2020 · standard

In the Solow growth model, the assumption of constant returns to scale means that…

Write what CRS says.
$F (λ K, λ L) = λ F (K, L)$
Choose λ = 1/L.
$F (\frac{K}{L}, 1) = \frac{1}{L} F (K, L) \Rightarrow y = f (k)$
Read off the meaning.
Output per worker depends on capital per worker alone. The number of workers does not enter.

Answer

(A). The number of workers does not affect the relationship between output per worker and capital per worker. (B) is false since countries differ in k; (C) confuses output per worker with total output; (D) mixes up two unrelated assumptions.

A harder one, for later

IIT JAM 2023 asks for the per-capita growth rate when Y = AK^αL^1−α, expressed using the capital–output ratio κ = K/Y rather than capital per worker. Divide by L: y = A(K/L)^α. Write K/L = κy and y = A(κy)^α, so y^1−α = Aκ^α, which is y = A^1/(1−α)κ^α/(1−α). Taking growth rates:

\frac{y ˙}{y} = \frac{1}{1 - α} \frac{A ˙}{A} + \frac{α}{1 - α} \frac{κ ˙}{κ}

That is option (A). The form says something useful: if the capital–output ratio is constant, as Kaldor’s fourth fact says, then per-worker growth comes entirely from A, magnified by 1/(1 − α).

Watch out for

The mistakes examiners are fishing for

Returns to scale against diminishing returns. The first scales all inputs, the second one input. Cobb–Douglas with a + b = 1 has both.
Assuming Cobb–Douglas is always CRS. It is CRS only if the exponents add to 1. Check before using constant returns.
MP equals AP. In Cobb–Douglas, MPK = α · APK. They coincide only when α = 1.
Mixing up the neutralities. Harrod: A multiplies L. Solow: A multiplies K. Hicks: A multiplies the whole function. Only Harrod is consistent with steady growth in general.
Shares that do not add up. Factor shares equal the exponents and sum to 1 only under constant returns and competitive pay.

Revise

The lesson on one screen

Production function: Y = F(K, L, A). Positive, diminishing marginal products.
Returns to scale: F(λK, λL) = λ^hF. Cobb–Douglas: h = a + b. Constant returns: h = 1.
Euler's theorem: For degree 1: K·F_K + L·F_L = F, so shares add to 1.
Cobb–Douglas: Y = AK^αL^1−α; MPK = αY/K; capital share α; g_Y = g_A + αg_K + (1 − α)g_L.
Intensive form: y = f(k) = Ak^α, with f′ > 0, f″ < 0 and Inada conditions. Needs CRS.
Neutral technical change: Hicks AF(K, L); Harrod F(K, AL); Solow F(AK, L). Balanced growth needs Harrod, except in Cobb–Douglas.

Practice

Check yourself

0 of 9 correct so far

A. Numbers

Problem 1

Q = K^0.3L^0.5. Both inputs are multiplied by 4. By what factor does output rise?

Factor

Problem 2

Y = 2K^0.5L^0.5, with K = 16 and L = 25. Find output, and the marginal product of capital.

YMPK

Problem 3

y = k^0.5. An economy has 900 units of capital and 100 workers. Find output per worker and total output.

B. Concepts

1Which statement is true for a production function homogeneous of degree one?

2Technical progress that makes each worker more effective, so Y = F(K, AL), is called

3In Y = AK^α L^(1−α), with inputs paid their marginal products, capital's share of income is

4Which Inada condition fails for Y = AK?

Real questions

Past papers, by level

Questions on homogeneity, marginal products, returns to scale and neutrality, from CUET-PG, Hyderabad, IIT JAM and UGC NET.

Foundation

Definitions, one-step arithmetic and which-model questions. Most CUET and Hyderabad questions sit here.

2 questionsPractise →

Standard

Growth accounting, steady-state algebra, statement questions and matching, up to JNU, IIT JAM and UGC NET level.

8 questionsPractise →

Delhi School level

This topic’s hardest questions, from the DSE entrance, are worked through in Lesson 14: Linear technology and the Inada conditions.

4 questionsLesson 14 →

All 10 past-paper questions for this lesson, easiest first: practise the full set →

These are questions from the TestStop bank. A free account answers 10 a day, each with its worked solution; All-Access opens every set with no daily limit.

Quick answers

Frequently asked questions

What is a production function?

A production function says how much output an economy or firm can produce from given amounts of inputs, usually capital K and labour L, and a level of technology A. In growth theory it is written Y = F(K, L, A), and the standard example is the Cobb–Douglas function Y = A K^α L^(1−α).

What are constant returns to scale, and why does growth theory assume them?

Constant returns to scale means that multiplying all inputs by the same factor multiplies output by that factor: double K and L and Y doubles. Growth theory assumes it because an economy can be replicated: two identical economies produce twice as much. It also lets output per worker depend only on capital per worker, y = f(k), which is what makes the Solow model solvable.

What is the Cobb–Douglas production function?

Y = A K^α L^β. The exponents are the output elasticities of capital and labour. If α + β = 1 there are constant returns to scale, and if inputs are paid their marginal products, capital receives a share α of income and labour a share 1 − α. Cobb and Douglas fitted it to US manufacturing data in 1928.

What are the Inada conditions?

They are limits on the shape of the production function that guarantee a steady state exists: output is zero with no capital, the marginal product of capital is positive and falling, it is infinite as capital approaches zero and goes to zero as capital becomes very large. Cobb–Douglas with capital's exponent between 0 and 1 satisfies them.

What is the difference between Hicks-neutral, Harrod-neutral and Solow-neutral technical change?

Hicks-neutral change multiplies the whole function, Y = A F(K, L), and leaves the ratio of marginal products unchanged at a given capital–labour ratio. Harrod-neutral (labour-augmenting) change makes labour more effective, Y = F(K, AL), and keeps the capital–output ratio constant at a given interest rate. Solow-neutral (capital-augmenting) change makes capital more effective, Y = F(AK, L). For a Cobb–Douglas function all three are the same thing written differently.

Why must technical progress be labour-augmenting for steady growth?

Uzawa (1961) showed that a neoclassical model can have a balanced growth path, with all variables growing at constant rates, only if technical progress is labour-augmenting (Harrod-neutral), apart from the Cobb–Douglas case where the forms coincide. That is why the Solow model with technical progress writes output as F(K, AL).

← Lesson 1What is economic growth? The facts, the rule of 70 and Kaldor's facts Lesson 3 →Growth accounting and the Solow residual

The production function: returns to scale, Cobb–Douglas and technical change

Output from inputs, and diminishing returns

Returns to scale

Euler’s theorem, and why constant returns are assumed

Cobb–Douglas and factor shares

The intensive form: output per worker

What shape should f(k) have?

Technical change and neutrality

The opening question, solved

A harder one, for later

The mistakes examiners are fishing for

The lesson on one screen

Check yourself

A. Numbers

B. Concepts

Past papers, by level

Foundation

Standard

Delhi School level

Frequently asked questions

Read next

The production function: returns to scale, Cobb–Douglas and technical change

Output from inputs, and diminishing returns

Returns to scale

Euler’s theorem, and why constant returns are assumed

Cobb–Douglas and factor shares

The intensive form: output per worker

What shape should f(k) have?

Technical change and neutrality

The opening question, solved

A harder one, for later

The mistakes examiners are fishing for

The lesson on one screen

Check yourself

A. Numbers

B. Concepts

Past papers, by level

Foundation

Standard

Delhi School level

Frequently asked questions

Read next