Differential Calculus and Marginal Analysis

Introduction

Business decisions often depend not only on current values but also on rates of change. A manager may ask: How rapidly do sales change when price changes? How much additional cost arises from producing one more unit? At what output level is profit maximised? Differential calculus provides a systematic language for answering these questions.

In business analytics, calculus is used to analyse demand, revenue, cost, profit, growth, optimisation and responsiveness. This chapter develops the essential single-variable tools: limits, derivatives, derivative rules, marginal functions, elasticity, stationary points and optimisation. The emphasis is on interpretation as well as calculation.

Learning objectives

By the end of this chapter, you should be able to:

• evaluate limits using algebraic and standard limit rules;
• explain the derivative as a limit, slope and instantaneous rate of change;
• differentiate common business functions using standard rules;
• interpret marginal cost, revenue and profit in context;
• calculate and interpret point elasticity;
• identify stationary points and classify local maxima and minima;
• solve single-variable business optimisation problems;
• recognise the assumptions and limitations of marginal analysis.

1. Limits and continuity

A limit describes the value that a function approaches as its input approaches a specified value. If f(x) approaches L as x approaches a, we write:

Here, x is the independent variable, a is the point being approached, and L is the limiting value. The function need not actually be defined at a for the limit to exist.

Limits are important because the derivative is defined using a limit. They also help analysts examine behaviour near thresholds, production levels, prices and capacity constraints.

1.1 Standard limit laws

When the individual limits exist, the following rules apply:

The quotient rule requires lim_x→ag(x) ≠ 0. For a polynomial, direct substitution is valid because polynomials are continuous everywhere. For a rational function, direct substitution is valid wherever the denominator is non-zero.

Worked example 1: A polynomial limit

Let f(x)=3x²−2x+5. Evaluate lim_x→2f(x). Since the function is a polynomial, substitute x=2:

Worked example 2: Removing a factor

Evaluate:

Direct substitution gives 0/0, which is indeterminate. Factor the numerator:

Therefore:

The original function may be undefined at x=3, but its limiting value is 6.

1.2 Continuity

A function is continuous at x=a when three conditions hold:

1. f(a) exists;
2. lim_x→af(x) exists;
3. lim_x→af(x)=f(a).

Continuous models are useful for quantities such as cost and revenue when small changes in output produce small changes in value. Discontinuities may represent fixed charges, capacity jumps, tax thresholds or changes in pricing policy.

2. The derivative and rates of change

The derivative measures instantaneous change. For a function y=f(x), its derivative at x is defined as:

The symbol h represents a small change in x. The derivative can be interpreted as:

• the slope of the tangent to the graph;
• the instantaneous rate of change of y with respect to x;
• the marginal effect of a small change in a business variable.

If revenue is measured in dollars and output in units, then R′(q) has units of dollars per unit. If q is measured in thousands of units, the derivative has units of dollars per thousand units unless the model is rescaled.

The average rate of change between x=a and x=b is:

The derivative is the limiting value of this average rate as the interval becomes infinitesimally small.

Worked example 3: Derivative from first principles

Let f(x)=x². Using the definition:

Expand and simplify:

Thus the instantaneous rate of change of x² is 2x.

3. Standard differentiation rules

The following rules assume that the relevant derivatives exist.

The last rule is the chain rule. It is particularly useful when a business model contains a function inside another function.

Important elementary derivatives include:

Worked example 4: Differentiating a cost function

Suppose total cost, in dollars, is:

where q is output in units. Differentiate term by term:

The fixed cost 500 disappears because its derivative is zero. The marginal cost function is therefore MC(q)=C′(q).

Worked example 5: Product and chain rules

Let demand-related revenue be:

Using the product rule with u=q and v=(100−2q)², and the chain rule for v,

Thus:

This can be simplified if required. The product and chain rules avoid treating the changing factors as constants.

4. Marginal analysis in business

Let q denote output or sales quantity. Define:

• C(q): total cost;
• R(q): total revenue;
• P(q): total profit.

The fundamental identities are:

and, when the functions are differentiable,

Marginal cost, marginal revenue and marginal profit are:

They measure the approximate change in the corresponding total quantity caused by a one-unit increase in output. For a small change ∆q,

This approximation is most reliable when ∆q is small and the function is smooth near q. It is not an exact statement for a large change.

Average versus marginal quantities

Average cost is:

Average revenue is:

Average profit is P(q)/q. Marginal cost is not generally equal to average cost. However, where the relevant functions are differentiable, marginal cost intersects average cost at an interior minimum of average cost. This follows because:

At a stationary point of average cost, AC′(q)=0, so C′(q)=C(q)/q, or MC=AC.

Revenue under a linear demand function

Suppose inverse demand is:

where p is price per unit. Revenue is price multiplied by quantity:

Therefore:

The feasible economic domain normally requires q≥0 and p(q)≥0, which implies 0≤q≤a/b.

5. Elasticity of demand

Elasticity measures proportional responsiveness rather than absolute change. If quantity demanded is Q(p), point price elasticity of demand is:

The usual sign is negative because demand often decreases as price increases. Analysts frequently use the absolute value |E_d| for classification:

• elastic: |E_d|>1;
• unit elastic: |E_d|=1;
• inelastic: |E_d|<1.

Elasticity is dimensionless. It approximates the percentage change in quantity resulting from a 1% change in price:

This point approximation assumes a sufficiently small price change and differentiability near the specified price. For a large change, arc elasticity is often preferable:

Worked example 6: Elasticity

Let Q(p)=1200−20p. At p=30, quantity is Q=600, and Q′(p)=−20. Hence:

Demand is unit elastic at this price. A small 1% increase in price is associated with approximately a 1% decrease in quantity demanded.

For a linear demand curve, total revenue is maximised where demand is unit elastic, provided the price and quantity are in the interior of the feasible domain. This result should not be applied blindly when demand is discontinuous, constrained by capacity or affected by strategic reactions.

6. Stationary points and optimisation

A stationary point occurs where the first derivative is zero:

It may be a local maximum, local minimum or neither. A zero derivative alone does not establish optimality.

First-derivative test

If f′(x) changes from positive to negative at x=c, the function has a local maximum. If it changes from negative to positive, the function has a local minimum. If there is no sign change, the stationary point may be stationary but not an extremum.

Second-derivative test

If f′(c)=0 and:

then f has a local maximum at c. If:

then it has a local minimum. If f″(c)=0, the test is inconclusive and the first-derivative test or another method is needed.

A global optimum on a closed interval requires comparison of all relevant stationary points and endpoints. In business, feasibility constraints must also be checked: negative output, negative prices or production beyond capacity may be mathematically possible but economically meaningless.

Profit maximisation

An interior profit maximum satisfies:

Since P′(q)=MR(q)−MC(q), this is equivalent to:

The condition is necessary for an interior optimum, not always sufficient. A maximum also requires an appropriate derivative sign change or curvature condition, and the result must satisfy all constraints.

Worked example 7: Profit optimisation

A firm faces inverse demand:

and total cost:

with q≥0. Revenue is:

Profit is:

Differentiate:

Set the derivative equal to zero:

The second derivative is:

Therefore, profit is maximised at 18 units. Price is:

Revenue is 62×18=1116, cost is 200+8(18)+18²=668, and maximum profit is 1116−668=448. The result also satisfies the feasible price condition because price is positive.

7. A systematic optimisation workflow

A robust business optimisation procedure is:

1. Define the decision variable and its units.
2. State the feasible domain and constraints.
3. Build the objective function, such as revenue, cost, profit or average cost.
4. Differentiate the objective.
5. Find all stationary points in the feasible domain.
6. Check second-derivative or first-derivative conditions.
7. Evaluate endpoints and boundary constraints.
8. Interpret the result in operational, financial and managerial terms.
9. Test sensitivity to assumptions and parameter changes.

This procedure prevents a common error: reporting a stationary point without demonstrating that it is feasible and optimal.

8. Illustrative case: pricing a subscription service

Background

The fictional firm StreamWise sells a monthly digital subscription. Its analyst estimates monthly demand as:

where p is monthly price in dollars and q is subscriptions. Variable and fixed costs are estimated by:

The model is illustrative and assumes stable customer behaviour, a continuous demand function and no competitor response.

Problem

Management wants to select a price that maximises monthly profit, subject to p≥0 and q≥0.

Analysis and solution

Solve the demand equation for price:

Revenue is:

Profit is:

Differentiate and solve:

The second derivative is P″(q)=−0.014<0, so this is a local maximum. Because the profit function is a concave quadratic and the feasible domain is an interval, it is also the global maximum within the model.

The corresponding price is:

Approximate maximum profit is:

Interpretation

The result means that, under the stated assumptions, a price of approximately $35.71 produces approximately 2,857 subscriptions and the highest estimated monthly profit. It does not prove that management should immediately charge that price. The estimate should be tested against customer research, capacity, churn, competitor responses, taxes, payment costs and uncertainty in the demand coefficients.

Learning points

The case demonstrates the relationship between inverse demand, revenue, cost and profit. It also shows why optimisation requires more than differentiation: the analyst must define units, check the domain, verify curvature and communicate assumptions.

9. Limitations, governance and professional practice

Calculus models are simplifications. Several limitations deserve attention.

Parameter uncertainty. Demand and cost coefficients are estimated from data and may be unstable. Confidence intervals, scenario analysis and sensitivity analysis should accompany point estimates.

Discrete decisions. A model may produce q=2,857.14, but subscriptions, vehicles and production lots may be integer or batch quantities. The nearest feasible values should be compared directly.

Non-smooth behaviour. Discounts, tax brackets, capacity thresholds and setup costs can create discontinuities or kinks. Ordinary derivative conditions may not apply at those points; boundary and one-sided analyses are required.

Causal interpretation. A derivative estimated from observational data describes an association unless the underlying demand or cost relationship has a defensible causal interpretation.

Strategic response. Competitors, suppliers and customers may react to a price or output decision. A static single-firm model may then overstate expected profit.

Ethics and governance. Pricing optimisation should be reviewed for discriminatory effects, exploitative practices and compliance with competition and consumer-protection law. Personal data used to estimate demand must be collected and processed lawfully, securely and transparently. Automated recommendations should be auditable, monitored for drift and subject to human accountability.

Technology and AI. Software such as spreadsheets, computer algebra systems and Python can calculate derivatives, plot functions and perform sensitivity analysis. These tools support—not replace—model specification and interpretation. AI-generated calculations should be independently verified, especially where rounding, units, constraints or hidden assumptions affect decisions.

Best practices for analysts

• Use explicit variable definitions and units.
• Separate mathematical feasibility from commercial feasibility.
• Report both the optimum and the assumptions producing it.
• Check derivatives symbolically and numerically where possible.
• Compare stationary points with endpoints and discrete alternatives.
• Use graphs and sensitivity analysis to communicate robustness.
• Avoid false precision: estimated prices should not be reported to more decimal places than the data justify.
• Validate the model with historical, experimental or pilot data.

Chapter Summary

Limits describe the value a function approaches and provide the foundation for the derivative. The derivative is an instantaneous rate of change and can be interpreted as the slope of a tangent or the marginal effect of a small change. Standard rules—including the power, product, quotient and chain rules—allow business functions to be differentiated efficiently.

For total cost, revenue and profit, the derivatives are marginal cost, marginal revenue and marginal profit. Profit satisfies P=R−C and, at an interior optimum, MR=MC, subject to feasibility and a maximum condition. Elasticity measures percentage responsiveness and is especially useful for interpreting pricing and revenue decisions. Stationary points are found by solving f′(x)=0, but classification requires derivative tests and comparison with boundaries.

Effective marginal analysis combines calculus with domain knowledge, data quality assessment, sensitivity analysis, ethical governance and clear communication. A mathematically optimal result is only as useful as the assumptions, measurements and institutional context supporting it.

Key Terms

Limit: The value a function approaches as its input approaches a specified point.

Continuity: A property in which the function value equals its limit at a point.

Derivative: The instantaneous rate of change of a function.

Marginal analysis: The study of how a small change in one variable affects an outcome.

Marginal cost: The derivative of total cost with respect to output.

Marginal revenue: The derivative of total revenue with respect to output.

Marginal profit: The derivative of total profit with respect to output.

Average cost: Total cost divided by positive output.

Elasticity: A dimensionless measure of proportional responsiveness.

Point elasticity: Elasticity evaluated at a particular price or quantity.

Arc elasticity: Elasticity calculated over a finite interval.

Stationary point: A point where the first derivative is zero.

Local maximum: A value at least as large as nearby function values.

Local minimum: A value no greater than nearby function values.

Global maximum: The greatest function value over the entire feasible domain.

Second-derivative test: A method using curvature to classify a stationary point.

Inverse demand: Price expressed as a function of quantity.

Feasible domain: Values satisfying the model’s mathematical and practical constraints.

Sensitivity analysis: Examination of how results change when assumptions or parameters change.

Review Questions

Short Answer Questions

1. What does lim_x→af(x) represent?
2. Define the derivative in terms of a limit.
3. Distinguish marginal cost from average cost.
4. What does price elasticity of demand measure?
5. Why must endpoints be checked in a constrained optimisation problem?

Long Answer Questions

1. Explain the economic interpretation of derivatives of cost, revenue and profit, including units and approximation conditions.
2. Describe the power, product, quotient and chain rules and illustrate when each may be used.
3. Explain how a firm can use MR=MC to locate an interior profit optimum, and why the condition alone is insufficient.
4. Compare point elasticity and arc elasticity, including the circumstances in which each is appropriate.
5. Discuss the limitations and governance risks of using calculus-based pricing models in a data-driven organisation.

Analytical / Application Questions

1. A firm has R(q)=120q−q² and C(q)=300+20q+0.5q². Find the profit-maximising output, verify the second-order condition and calculate the associated price if inverse demand is p(q)=120−q.
2. Demand is Q(p)=2,000−25p. Calculate point elasticity at p=40 and interpret the result for a small price increase.
3. Average cost is AC(q)=600/q+10+0.02q, with q>0. Find the output that minimises average cost and explain why checking the domain is necessary.

Multiple Choice Questions with Answers

Question 1

For f(x)=3x²−2x+5, what is lim_x→2f(x)?

A. 9

B. 11

C. 13

D. 15

Correct Answer: C. 13

Question 2

If C(q)=500+12q+0.04q²+0.0002q³, which is the marginal cost function?

A. 500+12q+0.04q²

B. 12+0.08q+0.0006q²

C. 12+0.04q+0.0002q²

D. 500+0.08q+0.0006q²

Correct Answer: B. 12+0.08q+0.0006q²

Question 3

If R(q)=aq−bq², with positive constants a and b, marginal revenue is:

A. a−bq

B. aq−2b

C. a−2bq

D. 2aq−bq²

Correct Answer: C. a−2bq

Question 4

A stationary point satisfies which condition?

A. f(x)=0

B. f′(x)=0

C. f″(x)=0

D. x=0

Correct Answer: B. f′(x)=0

Question 5

If f′(c)=0 and f″(c)<0, then c is generally a:

A. Local maximum

B. Local minimum

C. Discontinuity

D. Point of inflection necessarily

Correct Answer: A. Local maximum

Question 6

For Q(p)=1200−20p, what is point elasticity at p=30?

A. −2

B. −1

C. 1

D. 2

Correct Answer: B. −1

Question 7

If profit is P(q)=R(q)−C(q), which identity is correct?

A. P′(q)=R′(q)+C′(q)

B. P′(q)=C′(q)−R′(q)

C. P′(q)=R′(q)−C′(q)

D. P′(q)=R(q)−C(q)

Correct Answer: C. P′(q)=R′(q)−C′(q)

Question 8

For inverse demand p(q)=80−q, revenue at q=18 is:

A. $62

B. $80

C. $1,116

D. $1,440

Correct Answer: C. $1,116

Question 9

Which condition is required when applying the quotient limit rule?

A. The numerator must be zero

B. The denominator’s limit must be non-zero

C. The function must be quadratic

D. The variable must be an integer

Correct Answer: B. The denominator’s limit must be non-zero

Question 10

Why should endpoints be evaluated in a constrained optimisation problem?

A. Derivatives are never useful at stationary points

B. An endpoint may provide the global optimum even without a zero derivative

C. Endpoints always have zero profit

D. Endpoints make elasticity dimensionless

Correct Answer: B. An endpoint may provide the global optimum even without a zero derivative