Bachelors Program in Human Resource Management
Business Mathematics & Statistics
•
Bachelor
Calculus and Business Optimization
Introduction
Calculus provides methods for analysing change, accumulation and optimal decisions. In business, managers rarely choose among outcomes that remain constant. Labour costs change with staffing levels, sales respond to price, productivity changes with training investment, and employee turnover accumulates over time. Calculus converts such relationships into models that support disciplined decisions.
For human resource management (HRM), relevant applications include optimizing workforce size, allocating training budgets, estimating marginal labour cost, evaluating recruitment intensity and forecasting cumulative absenteeism. Calculus does not replace managerial judgement: its conclusions depend on the quality, units and assumptions of the underlying model.
Learning Objectives
After studying this chapter, students should be able to:
explain functions, limits, derivatives and integrals in business contexts;
calculate derivatives using standard differentiation rules;
interpret marginal cost, marginal revenue and marginal productivity;
distinguish stationary points from verified optima;
solve unconstrained and constrained optimization problems;
apply partial derivatives and Lagrange multipliers;
use integration to measure cumulative business quantities;
formulate dimensionally consistent HR optimization models; and
evaluate the practical, ethical and statistical limitations of calculus-based decisions.
1. Business Functions and Rates of Change
A function assigns one output to each admissible input. If training expenditure determines an estimated productivity gain, the relationship may be written g = f(x), where x is expenditure and g is productivity gain.
A model should identify:
the independent variable;
the dependent variable;
the feasible domain;
units of measurement; and
assumptions about other factors held constant.
1.1 Common functional forms
Form
General expression
Typical application
Linear
y = a + bx
Constant marginal labour cost
Quadratic
y = a + bx + cx2
Profit with diminishing returns
Exponential
y = Aekt
Continuous growth or decline
Logarithmic
y = a + b\ln x
Rapid early gains that diminish
Power
y = Axb
Production or learning relationships
Parameters must be interpreted with their units. In C(q)=F+cq, for example, C may be dollars, q units produced, F dollars and c dollars per unit.
1.2 Average and instantaneous change
For a change from x to x+\Delta x, the average rate of change is
The derivative is the limiting instantaneous rate of change:
f'(x)=\limh\to 0\frac{f(x+h)-f(x)}{h}.
Here, h has the same units as x. The derivative has units of output per unit of input. For example, if cost is measured in dollars and output in units, C'(q) is measured in dollars per additional unit.
Differentiability implies continuity at the point, but continuity does not necessarily imply differentiability. A function with a sharp corner may be continuous yet lack a derivative there.
2. Differentiation Rules
For differentiable functions f and g, and constant k, the principal rules are:
where p is dollars per unit and q is units sold. Its cost function is
C(q)=2{,}000+20q+0.05q2.
Revenue is price multiplied by quantity:
R(q)=p(q)q=120q-0.2q2.
Profit is
\Pi(q)=R(q)-C(q)=-0.25q2+100q-2{,}000.
The first derivative is
\Pi'(q)=-0.5q+100.
Setting \Pi'(q)=0 gives q=200. Because
\Pi''(q)=-0.5<0,
the profit function is strictly concave and the stationary point is the unique maximum. At this output,
p(200)=80, \qquad \Pi(200)=\$8{,}000.
Equivalently, marginal revenue and marginal cost are equal:
MR=120-0.4q, \qquad MC=20+0.1q.
At q=200, both equal $40 per unit.
3. Marginal Analysis
The derivative of a total function is its marginal function.
Total function
Derivative
Interpretation
C(q)
MC=C'(q)
Approximate cost of one more unit
R(q)
MR=R'(q)
Approximate revenue from one more unit
\Pi(q)
\Pi'(q)
Marginal profit
Q(L)
MPL=Q'(L)
Marginal product of labour
For a small change \Delta x, the differential approximation is
\Delta f \approx f'(x)\Delta x.
This is most reliable when the change is small and the function is smooth near x. A derivative is not always the exact effect of adding one indivisible worker or machine.
3.1 Revenue and price elasticity
If demand is q=q(p), point price elasticity is
Ep=\frac{dq}{dp}\frac{p}{q}.
Both p and q must be positive, and the demand function must be differentiable at the evaluated price. Ordinary downward-sloping demand produces Ep<0. Demand is elastic when |Ep|>1, unit elastic when |Ep|=1 and inelastic when |Ep|<1.
Revenue is R(p)=p q(p), so
\frac{dR}{dp}=q+p\frac{dq}{dp}=q(1+Ep).
This identity applies only where q>0 and the derivative exists.
Worked Example 2: An indivisible decision
An estimated annual profit model for a service team is
\Pi(q)=-500q2+133{,}733q-8{,}916{,}178,
where q is an integer number of service packages and profit is dollars. The continuous stationary point is
q^*=\frac{133{,}733}{1{,}000}=133.733.
Because packages are indivisible, nearby feasible integers must be evaluated:
\Pi(133)=\$25{,}811,
\Pi(134)=\$26{,}044.
Thus q=134 is better than q=133. Also, \Pi(135)=\$25{,}277, confirming that 134 is the maximizing feasible integer. Rounding a continuous optimum without checking neighbouring feasible values can produce an incorrect decision.
Worked Example 3: Elasticity of training demand
Suppose demand for external training places is
q(p)=1{,}000-20p,
where p is the fee in dollars and q is places demanded. At p=30, demand is 400 places and
Ep=(-20)\frac{30}{400}=-1.5.
Demand is elastic. Moreover,
\frac{dR}{dp}=400+30(-20)=-200.
Near $30, a $1 increase in the fee is estimated to reduce revenue by approximately $200.
4. Unconstrained Optimization
Optimization identifies the best feasible value of an objective function. A maximum may represent highest profit, productivity or service value; a minimum may represent lowest cost, waiting time or turnover.
4.1 Necessary and sufficient conditions
If f is differentiable and has an interior local optimum at x^*, Fermat’s theorem gives the necessary condition
f'(x^*)=0.
A solution of this equation is a stationary point, not automatically an optimum.
If f'(x^*)=0, the second-derivative test states:
f''(x^*)<0 implies a strict local maximum;
f''(x^*)>0 implies a strict local minimum;
f''(x^*)=0 is inconclusive.
For a global decision on a closed interval [a,b], evaluate all interior critical points, nondifferentiable points and endpoints. The Extreme Value Theorem guarantees that a continuous function on a closed and bounded interval attains a maximum and minimum.
If f is concave over the entire feasible interval, every local maximum is global. Strict concavity gives at most one maximizer.
Concave profit function with a unique maximum
Figure: Concave profit function with a unique maximum
Figure: A strictly concave objective has a unique interior maximum when its first derivative is zero.
4.2 Boundary and nonsmooth solutions
Some optimum decisions occur at a boundary. If a recruitment campaign cannot exceed $50,000, the best feasible choice may be exactly $50,000 even though the unconstrained stationary point lies above it. Piecewise wage schedules, overtime premiums and fixed hiring costs can also create corners or discontinuities for which ordinary first-order conditions are insufficient.
5. Multivariable Optimization
Business outcomes usually depend on several inputs. If z=f(x,y), the partial derivative
\frac{\partial f}{\partial x}
measures the instantaneous change in z as x changes while y is held constant.
An unconstrained interior stationary point satisfies
For twice-differentiable functions, the Hessian matrix contains second-order partial derivatives. A negative-definite Hessian establishes strict local concavity and hence a strict local maximum. In two variables, sufficient conditions at a stationary point are fxx<0 and fxxfyy-(fxy)2>0.
6. Constrained Optimization and Lagrange Multipliers
Suppose the objective is f(x,y) and the equality constraint is g(x,y)=b. The Lagrangian is
These conditions require differentiability and a regular constraint, meaning that the gradient of the constraint is not zero at the solution. They identify candidates; concavity, convexity or second-order analysis is needed to establish optimality.
The multiplier \lambda is a shadow value. Under standard regularity conditions, it measures the approximate increase in optimized objective value from a one-unit increase in the constraint’s right-hand side b. Its units are objective units per resource unit.
Worked Example 4: Allocating a training budget
Let x and y denote thousands of dollars allocated to technical and leadership training. The available budget is $25,000, so
x+y=25.
Suppose the estimated benefit, measured in organizational benefit points, is
B(x,y)=13.476(x+y)-0.4x2-0.625y2.
The coefficient units are defined so that every term is measured in benefit points. The negative quadratic terms represent diminishing returns.
An additional $1,000 of budget would therefore increase the optimized objective by approximately 1.281 benefit points, provided the local model and active constraint remain valid. Because the Hessian is diagonal with negative entries -0.8 and -1.25, the benefit function is strictly concave and the solution is unique.
For inequality constraints, such as g(x,y)\le b, the broader Karush–Kuhn–Tucker conditions apply. These include primal feasibility, dual feasibility and complementary slackness. A nonbinding resource constraint normally has a zero shadow value.
7. Integration and Accumulation
Differentiation measures instantaneous change; integration measures accumulation. If r(t) is a rate per unit of time, then the total accumulated amount from t=a to t=b is
\intab r(t)\,dt.
The units confirm the interpretation. If r(t) is employees per month and t is months, the integral is employees over the interval.
The Fundamental Theorem of Calculus states that if F'(x)=f(x) and f is continuous, then
\intab f(x)\,dx=F(b)-F(a).
Suppose employee departures occur at the estimated rate
m(t)=12+0.5t
employees per month, where t is months after the start of the year. Estimated departures during the first six months are
For observed discrete monthly data, numerical integration or direct summation may be more appropriate than imposing a continuous curve.
8. Illustrative Case: Northbridge Workforce Service Capacity
Background
Northbridge Support Services is a fictional organization considering additional specialist staff time to reduce case delays. Management expresses additional staff time as s hundreds of staff-hours per month. Thus s=1 means 100 staff-hours.
Problem
Management wishes to balance service improvement against incremental monetary cost. Subtracting dollars directly from service-value points would be dimensionally invalid. Northbridge therefore uses a formally approved conversion factor of 2 benefit points per $1,000 of incremental cost. This factor reflects its documented budget-value framework rather than a universal economic constant.
Analysis
Estimated gross service value is
V(s)=70s-8s2
benefit points. Incremental cost is
C(s)=13.86s+2.25s2
thousands of dollars. Applying the conversion factor \kappa=2 benefit points per $1,000 gives a converted cost burden of 2C(s) benefit points. The dimensionally consistent net objective is therefore
N(s)=V(s)-\kappa C(s) =42.28s-12.5s2.
The feasible range is 0\le s\le 3, corresponding to zero to 300 additional staff-hours.
Differentiating,
N'(s)=42.28-25s.
The stationary point is
s^*=\frac{42.28}{25}=1.6912.
Because N''(s)=-25<0, the objective is strictly concave. The stationary point lies inside the feasible interval and is the unique maximum. It corresponds to approximately
100(1.6912)=169.12
additional staff-hours per month.
At the optimum, net benefit is approximately 35.752 benefit points. The estimated incremental monetary cost is approximately $29,875 per month before conversion into benefit points.
Approach and Outcome
Northbridge should use approximately 169 additional staff-hours if the fitted relationships, feasible interval and conversion factor are accepted. The result is not simply a financial profit maximum; it is a multi-criteria decision expressed in a common benefit-point scale.
Learning Points
Variables and coefficients must carry coherent units.
Dollars cannot be subtracted directly from service points without an explicit conversion.
A stationary point must be checked against curvature and boundaries.
The conversion factor is a governance choice and should undergo sensitivity analysis.
The mathematical optimum is an input to—not a substitute for—professional judgement.
9. Model Risk, Ethics and Governance
Calculus gives precise conclusions from specified functions, but estimated business functions are uncertain. Important limitations include:
Functional-form risk: A quadratic may fit historical data yet behave unrealistically outside the observed range.
Parameter uncertainty: Estimated coefficients have sampling error.
Causal ambiguity: Correlation between training and productivity may reflect selection effects.
Indivisibility: Employees, teams and facilities may not be continuously divisible.
Dynamic effects: A static optimum can ignore future learning, turnover or morale.
Multiple objectives: Cost, fairness, wellbeing and service quality may conflict.
Implementation constraints: Labour law, collective agreements and workload standards restrict feasible choices.
Ethically responsible HR optimization requires more than maximizing a numerical score. Decisions must not treat workers solely as adjustable inputs. Models should be audited for discriminatory effects, unsafe workload recommendations and biased historical data. Employee privacy must be protected when productivity, absence or performance data are used.
A well-governed optimization process should:
document variables, units and data sources;
state the feasible domain and legal constraints;
validate the estimated function out of sample;
conduct sensitivity and scenario analysis;
involve HR, finance, operations and affected employees;
maintain human review of high-impact decisions; and
monitor realized outcomes after implementation.
10. Technology and Contemporary Practice
Spreadsheet solvers, statistical software and programming languages can differentiate fitted functions, solve nonlinear models and visualize sensitivity. Automatic differentiation is increasingly used in machine learning and large-scale optimization. Digital workforce-planning systems can also update marginal estimates as new data arrive.
Technology does not guarantee validity. Software may return a stationary point outside the feasible domain or report a local optimum when several local solutions exist. Generative AI can assist with model formulation or code, but every equation, unit, constraint and numerical result requires independent verification. Reproducible calculations and documented assumptions are essential professional practices.
Chapter Summary
Calculus supports business decisions involving change, marginal effects, accumulation and optimization. Derivatives measure instantaneous rates such as marginal cost, revenue and labour productivity, while integrals measure cumulative quantities. Optimization requires more than solving a first-order equation: managers must evaluate curvature, boundaries, feasibility and indivisible alternatives. Partial derivatives extend marginal reasoning to several inputs, and Lagrange multipliers support constrained resource allocation while providing shadow-value interpretations. Reliable application requires dimensionally consistent objectives, validated assumptions, sensitivity analysis and ethical governance. In HRM, mathematical optimality must remain subordinate to legal obligations, employee wellbeing and informed managerial judgement.
Key Terms
Limit: The value approached by a function as its input approaches a specified point.
Derivative: The instantaneous rate of change of one variable with respect to another.
Partial derivative: The derivative with respect to one input while other inputs are held constant.
Marginal cost: The derivative of total cost with respect to output.
Marginal revenue: The derivative of total revenue with respect to quantity.
Marginal product: The additional output associated with a small increase in an input.
Elasticity: A unit-free measure of proportional responsiveness.
Stationary point: A differentiable point at which the first derivative or gradient is zero.
Critical point: A point where the derivative is zero or does not exist.
Concavity: Curvature associated with nonincreasing marginal values.
Convexity: Curvature associated with nondecreasing marginal values.
Objective function: The quantity to be maximized or minimized.
Constraint: A condition restricting feasible decisions.
Lagrangian: A function combining an objective and constraints through multipliers.
Shadow value: The approximate change in optimized value from relaxing a constraint by one unit.
Hessian matrix: A matrix of second-order partial derivatives.
Definite integral: The accumulated net quantity over an interval.
Sensitivity analysis: Examination of how conclusions change when assumptions or parameters vary.
Dimensional consistency: The requirement that quantities combined in an equation have compatible units.
Review Questions
Short Answer Questions
What is the difference between an average rate of change and a derivative?
Why is a stationary point not necessarily an optimum?
State the business interpretation of marginal cost.
What does a Lagrange multiplier represent?
Why must endpoints be evaluated in optimization over a closed interval?
Long Answer Questions
Explain how first- and second-order conditions are used in unconstrained optimization.
Discuss the relationship among demand elasticity, price changes and revenue.
Explain constrained optimization using the Lagrangian method, including its assumptions and limitations.
Discuss how integration can be applied to employee turnover, absenteeism or labour-cost analysis.
Critically evaluate the ethical and governance issues associated with workforce optimization models.
Analytical / Application Questions
A firm has R(q)=90q-0.3q2 and C(q)=1{,}500+18q+0.1q2. Determine the continuous profit-maximizing output, verify the second-order condition and calculate maximum profit.
An HR department allocates a $40,000 budget between recruitment and training. Formulate a suitable constrained optimization model, clearly defining variables, units, objective and assumptions.
An absence rate is a(t)=8+0.4t employee-days per month. Calculate cumulative absence over ten months and interpret the result’s units.
Multiple Choice Questions with Answers
Question 1
If total cost is C(q)=500+12q+0.2q2, what is marginal cost?
A. 12+0.4q
B. 500+12q
C. 12+0.2q
D. 500+0.4q
Correct Answer: A. 12+0.4q
Question 2
In Worked Example 1, what output maximizes continuous profit?
A. 80 units
B. 100 units
C. 200 units
D. 400 units
Correct Answer: C. 200 units
Question 3
For the indivisible profit model in Worked Example 2, which feasible quantity is optimal?
A. 132 packages
B. 133 packages
C. 134 packages
D. 135 packages
Correct Answer: C. 134 packages
Question 4
If price elasticity of demand is -1.5, demand is classified as:
A. Perfectly inelastic
B. Inelastic
C. Unit elastic
D. Elastic
Correct Answer: D. Elastic
Question 5
Under standard regularity conditions, a Lagrange multiplier measures:
A. The objective’s average value
B. The approximate optimized-value gain from one additional resource unit
C. The total fixed cost of the constraint
D. The probability that the constraint binds
Correct Answer: B. The approximate optimized-value gain from one additional resource unit
Question 6
In Worked Example 4, approximately how much is allocated to technical training?
A. $9,756
B. $12,500
C. $15,244
D. $25,000
Correct Answer: C. $15,244
Question 7
If departures occur at m(t)=12+0.5t employees per month, how many departures are estimated over the first six months?
A. 72
B. 75
C. 81
D. 90
Correct Answer: C. 81
Question 8
Which condition is sufficient for a stationary point of a twice-differentiable one-variable function to be a strict local maximum?
A. f''(x^*)<0
B. f''(x^*)>0
C. f'(x^*)>0
D. f(x^*)=0
Correct Answer: A. f''(x^*)<0
Question 9
In the Northbridge case, the optimal additional staff time is approximately:
A. 125 staff-hours
B. 150 staff-hours
C. 169 staff-hours
D. 300 staff-hours
Correct Answer: C. 169 staff-hours
Question 10
Why is a conversion factor introduced in the Northbridge objective?
A. To make the objective linear
B. To guarantee that monetary costs are zero
C. To express service value and monetary cost in compatible units
D. To remove the feasible-capacity constraint
Correct Answer: C. To express service value and monetary cost in compatible units