Source: Lecture 10 – Production
Difficulty: Intermediate | Prerequisites: Consumer theory (utility maximisation, budget constraints, indifference curves, MRS).
This lecture marks the pivot from consumer theory to producer theory, the other half of microeconomics. Everything you learnt about how consumers maximise utility under a budget constraint has a direct parallel on the firm side: firms maximise profit under a cost constraint, using a production function instead of a utility function.
If you are comfortable with indifference curves, the marginal rate of substitution, and diminishing marginal utility, you already have the conceptual scaffolding. The new vocabulary (isoquants, marginal product, technical rate of substitution) maps almost one-to-one onto those consumer-side tools.
You should already know how to take partial derivatives and interpret them as marginal changes, because that is the main mathematical operation throughout this material.
Firms combine inputs (capital, labour, land, raw materials) through a production function to produce output. The Cobb-Douglas function f(K, L) = AK^a L^b is the workhorse model. Each input has a marginal product (extra output from one more unit), and the ratio of marginal products gives the technical rate of substitution, which measures how easily the firm can swap one input for another while holding output constant. Marginal products diminish as you pile on more of one input, and whether output scales proportionally, more than proportionally, or less depends on the sum of the exponents.
Theory of the firm
The branch of microeconomics that studies how producers make decisions about what to produce and how to produce it. Think of it as the mirror image of consumer theory, but with profit in place of utility.
Factors of production (inputs)
The resources a firm uses to make its product: capital, labour, land, and raw materials. In simple terms, these are the ingredients in the firm's recipe.
Capital (K)
Inputs that are themselves produced goods, such as machinery, buildings, computers, and software. Sometimes split into physical capital (tangible equipment) and financial capital (money used to start or run the business).
Labour (L)
The human work required to produce goods. In models, usually measured as hours of work per time period.
Production function
A mathematical expression that maps a set of inputs to the maximum output achievable with those inputs. Think of it as the firm's recipe: given these ingredients in these quantities, here is the most you can produce.
Cobb-Douglas production function
The specific functional form f(K, L) = AK^a L^b, where A is total factor productivity, K is capital, L is labour, and a and b are the output elasticities with respect to each input. It is the standard workhorse model in intermediate micro.
Total factor productivity (TFP, A)
The scaling parameter in the Cobb-Douglas function. It captures the overall efficiency of the production process, including technology and organisation. In simple terms, it is how much output you get from one unit of every input.
Isoquant
A curve showing all combinations of inputs that produce the same level of output. Think of it as the producer's version of an indifference curve.
Marginal product (MP)
The additional output produced by one more unit of a single input, holding all other inputs constant. Formally, the partial derivative of the production function with respect to that input.
Technical rate of substitution (TRS)
The rate at which a firm can replace one input with another while keeping output unchanged. Calculated as the ratio of marginal products: TRS = MP_K / MP_L. This is the producer-side equivalent of the marginal rate of substitution (MRS) from consumer theory.
Law of diminishing marginal product
The observation that, holding other inputs fixed, each additional unit of one input eventually adds less and less to total output. In simple terms, the tenth worker in a one-room factory adds less output than the second worker did.
Returns to scale
What happens to output when all inputs are scaled up by the same proportion. Constant returns to scale means output scales proportionally; increasing returns means output more than doubles when inputs double; decreasing returns means output less than doubles.
Short run vs. long run
In the short run, at least one input (typically capital) is fixed. In the long run, all inputs are variable. This distinction matters because diminishing marginal product bites hardest when some inputs cannot adjust.
Inputs are grouped into four broad categories:
Capital (K): produced goods used in production, such as machines, buildings, computers. Includes physical capital and financial capital (the money needed to acquire it).
Labour (L): human work, typically measured as hours per period.
Land: critical for sectors like agriculture, often folded into K for other industries.
Raw materials: unproduced natural inputs (ore, timber, water). Often bundled with capital in simple models, but modelled separately when analysing resource markets.
Inputs are treated as flows: a firm needs a certain number of labour-hours per week to produce a certain quantity of output per week.
The production function describes the maximum output obtainable from a given combination of inputs, assuming the firm uses the most technologically efficient method available.
A concrete example: producing one gallon of lemonade requires 3 cups of sugar, 10 lemons, 13 cups of water, 1 bag of ice, 1 hour of time, and 32 cups. Because every ingredient is needed in fixed proportions, the production function is a Leontief (fixed-proportions) function, where output equals the minimum of each input divided by its required quantity.
For most modelling purposes, inputs are aggregated into K and L, and the Cobb-Douglas form is used:
f(K, L) = A K^a L^b
Each parameter has a clear interpretation:
A (total factor productivity): the amount of output from one unit of each input. Captures technology and organisational efficiency.
K: quantity of capital.
L: quantity of labour.
a (capital's output elasticity): the responsiveness of output to changes in capital, or the share of output attributable to capital.
b (labour's output elasticity): the responsiveness of output to changes in labour, or the share of output attributable to labour.
The sum a + b determines returns to scale (covered below).
An isoquant is a curve plotting all input combinations (K, L) that yield the same output level. It is the producer's equivalent of the consumer's indifference curve.
The shape of an isoquant depends on the production function:
Cobb-Douglas (e.g. K^0.5 L^0.5): smooth, convex curves bowing toward the origin. Inputs are substitutable but not perfectly so.
Perfect substitutes (f = K + L): straight-line isoquants with slope -1. One unit of K always replaces one unit of L.
Perfect complements / Leontief (f = min{K, L}): L-shaped isoquants with a kink. Inputs must be used in fixed proportions; extra units of one input alone are worthless.
The marginal product of an input is the additional output from one more unit of that input, holding everything else constant. Formally, it is the partial derivative of the production function with respect to that input:
MP_K = ∂f(K, L) / ∂K
MP_L = ∂f(K, L) / ∂L
Worked example with f(K, L) = K^(1/2) L^(1/2):
MP_K = (1/2) K^(-1/2) L^(1/2) = (1/2)(L/K)^(1/2)
MP_L = (1/2) K^(1/2) L^(-1/2) = (1/2)(K/L)^(1/2)
Notice that as K rises (with L held constant), MP_K falls, and vice versa.
The TRS measures how much extra capital is needed to replace one unit of labour while keeping output the same. It is calculated as the (negative) ratio of marginal products:
TRS = MP_K / MP_L
(The negative sign is conventionally attached but often ignored, just as with the MRS in consumer theory.)
Worked example (continuing from above):
TRS = [(1/2)(L/K)^(1/2)] / [(1/2)(K/L)^(1/2)] = L / K
If L = K = 10, then TRS = 1. Losing one unit of labour requires exactly one additional unit of capital to maintain output.
Holding one input fixed, each successive unit of the other input produces less additional output. This is visible in the numerical table from the lecture: with K = 1 and L rising from 1 to 20, the marginal product of labour falls steadily from 1.000 to 0.113.
This pattern is sometimes called the "law" of diminishing marginal product, though it is an empirical regularity rather than a formal law. It reflects a common-sense constraint: cramming more and more workers into one factory (the "too many cooks" problem) yields smaller and smaller gains.
Diminishing marginal product also implies a diminishing TRS, which is why isoquants are convex to the origin.
Diminishing marginal product is especially relevant in the short run, when at least one input is fixed. Capital, for instance, cannot be adjusted quickly (building a factory takes time), so in the short run only labour is variable. In the long run, all inputs are adjustable.
Returns to scale ask a different question: what happens to output when all inputs are scaled by the same factor t > 1?
Constant returns to scale (CRS): f(tK, tL) = t f(K, L). Doubling inputs doubles output. For Cobb-Douglas, this holds when a + b = 1. Example: f = K^(1/2) L^(1/2).
Increasing returns to scale (IRS): f(tK, tL) > t f(K, L). Doubling inputs more than doubles output. Holds when a + b > 1. Example: f = K^2 L^2, where scaling by t multiplies output by t^4.
Decreasing returns to scale (DRS): f(tK, tL) < t f(K, L). Doubling inputs less than doubles output. Holds when a + b < 1. Example: f = K^(1/4) L^(1/4), where scaling by t multiplies output by t^(1/2).
The quick test for Cobb-Douglas: add the exponents. Sum equals 1 gives CRS, greater than 1 gives IRS, less than 1 gives DRS.
f(K, L) = A K^{a} L^{b}MP_K = \frac{\partial f(K,L)}{\partial K} \qquad MP_L = \frac{\partial f(K,L)}{\partial L}TRS = \frac{MP_K}{MP_L}For the specific case f(K, L) = K^(1/2) L^(1/2):
MP_K = \frac{1}{2}\left(\frac{L}{K}\right)^{1/2} \qquad MP_L = \frac{1}{2}\left(\frac{K}{L}\right)^{1/2} \qquad TRS = \frac{L}{K}Returns to scale test for Cobb-Douglas:
a + b = 1: constant returns to scale
a + b > 1: increasing returns to scale
a + b < 1: decreasing returns to scale
Leontief (fixed-proportions) production function:
f(x_1, x_2, \ldots, x_n) = \min\left\{\frac{x_1}{a_1}, \frac{x_2}{a_2}, \ldots, \frac{x_n}{a_n}\right\}where each a_i is the amount of input i required per unit of output.
Diminishing marginal product is why a restaurant cannot just keep hiring more waiters without also expanding its kitchen and floor space. At some point, the extra waiter is bumping into everyone else.
Returns to scale help explain why some industries tend toward a few large firms (increasing returns, where bigger is more efficient) and others sustain many small ones (decreasing returns, where growth hits a wall). Software, for example, often exhibits increasing returns to scale because the cost of producing one more copy is near zero.
Diminishing marginal product means total output is falling. It does not. Output is still rising; it is rising more slowly with each additional unit of the variable input. Output only falls if the marginal product goes negative, which is a separate (and unusual) case.
Returns to scale and diminishing marginal product are the same thing. They are not. Diminishing marginal product is about increasing one input while holding others fixed. Returns to scale is about scaling all inputs together by the same proportion. A production function can exhibit diminishing marginal product for each input individually while still having constant or even increasing returns to scale.
The TRS is always equal to L/K. Only for the specific Cobb-Douglas form f = K^(1/2) L^(1/2). For a general Cobb-Douglas f = AK^a L^b, TRS = (a/b)(L/K). Students often forget the coefficient (a/b) and apply the simplified result universally.
Isoquants can cross. They cannot. If two isoquants crossed, a single input combination would simultaneously produce two different output levels, which contradicts the definition of the production function as the maximum output from a given set of inputs.
⚠️ You will almost certainly be asked to compute marginal products and the TRS for a given Cobb-Douglas function. Practise the partial derivatives until they are automatic.
⚠️ Expect a question that asks you to determine returns to scale from a production function. The shortcut: add the exponents. Know the three cases cold.
⚠️ Be ready to distinguish between diminishing marginal product and decreasing returns to scale. Examiners like this because students consistently confuse the two.
⚠️ Drawing and interpreting isoquants for different production function types (Cobb-Douglas, perfect substitutes, Leontief) is standard exam material. Know the shapes.
True or false: If the marginal product of labour is diminishing, total output must be falling. False. Output is still increasing, just at a slower rate.
Fill in the blank: For a Cobb-Douglas function f = AK^a L^b, the production function exhibits constant returns to scale when a + b = 1.
True or false: An isoquant for a Leontief production function is L-shaped. True.
Fill in the blank: The technical rate of substitution is defined as the ratio of marginal products (MP_K / MP_L).
True or false: A production function with increasing returns to scale must also have increasing marginal products. False. Diminishing marginal products for individual inputs are compatible with increasing returns to scale when all inputs are scaled together.
Q: Given f(K, L) = 4K^(1/3) L^(2/3), calculate the marginal product of capital and the marginal product of labour.
A: MP_K = (4/3) K^(-2/3) L^(2/3) = (4/3)(L/K)^(2/3). MP_L = (8/3) K^(1/3) L^(-1/3) = (8/3)(K/L)^(1/3).
Q: For the same production function, what is the TRS?
A: TRS = MP_K / MP_L = [(4/3)(L/K)^(2/3)] / [(8/3)(K/L)^(1/3)] = (1/2)(L/K).
Q: Does f(K, L) = 4K^(1/3) L^(2/3) exhibit constant, increasing, or decreasing returns to scale?
A: The exponents sum to 1/3 + 2/3 = 1, so the function exhibits constant returns to scale.
Q: Suppose f(K, L) = K^(3/4) L^(3/4). Does this exhibit increasing, constant, or decreasing returns to scale? Prove it.
A: f(tK, tL) = (tK)^(3/4)(tL)^(3/4) = t^(3/2) K^(3/4) L^(3/4) = t^(3/2) f(K, L). Since 3/2 > 1 for any t > 1, t^(3/2) > t, so f(tK, tL) > t f(K, L). This is increasing returns to scale.
Q: Explain in words why diminishing marginal product of labour does not imply decreasing returns to scale.
A: Diminishing marginal product describes what happens when you add more of one input while holding the others fixed. Returns to scale describes what happens when you scale all inputs together. When both K and L increase proportionally, the constraint imposed by fixed capital is relieved, so the answer can differ. For example, f = K^(1/2) L^(1/2) has diminishing marginal product for each input individually, but constant returns to scale.
This material connects directly to cost theory (the next major topic in most intermediate micro courses), where the production function determines the shape of cost curves. A firm's cost-minimisation problem is the dual of the output-maximisation problem introduced here, and the TRS reappears as the condition for tangency between the isoquant and the isocost line.
The parallel to consumer theory is deliberate and examinable: utility functions map to production functions, indifference curves map to isoquants, MRS maps to TRS, and budget constraints map to isocost lines. If you can solve the consumer's problem, you can solve the producer's problem with the same techniques.
Production theory, theory of the firm, production function, Cobb-Douglas, factors of production, inputs and outputs, capital and labour, physical capital, financial capital, total factor productivity, TFP, isoquant, isoquant map, marginal product, marginal product of capital, marginal product of labour, MP_K, MP_L, technical rate of substitution, TRS, MRTS, marginal rate of technical substitution, diminishing marginal product, diminishing returns, law of diminishing returns, returns to scale, constant returns to scale, CRS, increasing returns to scale, IRS, decreasing returns to scale, DRS, economies of scale, short run, long run, fixed input, variable input, Leontief production function, fixed proportions, perfect substitutes in production, convex isoquant, ECN 312, intermediate microeconomics, producer theory