Mathematics for Economists (ECON613) sits at the heart of modern economic analysis: it equips you to model how agents behave, how markets clear, how data relate to theory, and how policies affect outcomes. This study guide builds the mathematical foundations typically assessed in ECON613 at South African universities, with worked examples, practice workflows, and exam-style checklists. Emphasis is placed on the math you actually use in economics—optimization, constrained choice, comparative statics, probability/statistics for econometrics, and linear algebra—and on how to present solutions clearly under time pressure.

1) Mathematical Foundations for Economic Modeling (Functions, Limits, and Rates)

ECON613 exams often begin by testing whether you can manipulate mathematical objects fluently: functions, graphs, limits, derivatives, integrals, and series. Economic models are written in compact mathematical language, and the difference between a strong and weak answer is usually not “knowing the topic,” but doing the algebra and interpretation correctly.

1.1 Functions, Notation, and Economic Interpretation

A function is a rule that maps inputs (variables) to outputs (outcomes). Economics uses functions constantly:

• Demand: (Q_d = D(P)) meaning quantity demanded depends on price (P).
• Cost: (C = C(q)) meaning cost depends on output (q).
• Production: (q = f(K,L)) with capital (K) and labour (L).

Key notation you should be able to use:

• (f(x)): value of function (f) at (x)
• (f'(x)): derivative (marginal effect)
• (\nabla f) or (\partial f/\partial x_i): gradients/partial derivatives in multivariable contexts
• (\int_a^b f(x),dx): area accumulation (often linked to total effect)

Economic meaning (the exam expects interpretation):

• If (y = f(x)) and (f'(x) > 0), then (y) increases as (x) increases.
• If (f'(x) < 0), then (y) decreases as (x) increases.
• If (f''(x) > 0), the slope is increasing (convexity), which matters for costs and utility.

Worked example: Marginal revenue from a revenue function

Suppose a firm’s revenue is
[
R(q)=50q-2q^2.
]
Then marginal revenue is
[
MR(q) = R'(q) = 50 – 4q.
]
At (q=10),
[
MR(10)=50-40=10.
]
Interpretation: increasing output from 10 by a small amount increases revenue by about (10) rand per unit (in a local sense). In an exam, you must state that link between marginal changes and the derivative.

1.2 Limits and Continuity (Model Stability)

Limits are used when models approach equilibrium or when functions are not defined at a point but have a meaningful limiting value.

You should know:

• How to evaluate simple limits (direct substitution, factorization)
• One-sided limits
• Continuity conditions: a function is continuous at (x=c) if
  1. (f(c)) exists,
  2. (\lim_{x\to c} f(x)) exists,
  3. (\lim_{x\to c} f(x)=f(c)).

Worked example: Limit with factorization

Evaluate:
[
\lim_{x\to 2}\frac{x^2-4}{x-2}.
]
Factor numerator:
[
x^2-4=(x-2)(x+2).
]
So:
[
\frac{(x-2)(x+2)}{x-2} = x+2 \quad (x\neq 2).
]
Thus the limit is
[
\lim_{x\to 2}(x+2)=4.
]

Exam tip: showing cancellation clearly earns marks; stating “use L’Hôpital’s rule” without checking alternatives is rarely needed for econ problems at first-year/mid-level.

1.3 Differentiation Rules and Economic Margins

You must be able to differentiate composite functions, products, quotients, and powers, including negative and fractional exponents.

Core rules:

• Power rule: (\frac{d}{dx}x^n = nx^{n-1})
• Product: ((uv)' = u'v + uv')
• Quotient: (\left(\frac{u}{v}\right)' = \frac{u'v – uv'u}{v^2})
• Chain rule: if (y=f(g(x))), then (y' = f'(g(x))g'(x))

Worked example: Chain rule with an economic utility function

Let utility be
[
U(c)=\ln(1+c^2).
]
Then
[
U'(c)=\frac{1}{1+c^2}\cdot 2c = \frac{2c}{1+c^2}.
]
Interpretation: this is the marginal utility of consumption (MU_c). It increases initially with (c) but eventually decreases because the denominator grows faster than numerator.

1.4 Higher-Order Derivatives: Concavity/Convexity

In economics, second derivatives are used to check curvature:

• Utility (U(x)) is often concave: (U''(x) < 0), implying diminishing marginal utility.
• Cost (C(q)) is often convex: (C''(q) > 0), implying increasing marginal cost.
• Production functions may have special concavity properties.

Worked example: Concavity test

Suppose
[
U(x) = 10\sqrt{x} = 10x^{1/2}.
]
Then
[
U'(x)=10\cdot \frac{1}{2}x^{-1/2} = 5x^{-1/2},
]
and
[
U''(x)=5\cdot (-\frac{1}{2})x^{-3/2} = -\frac{5}{2}x^{-3/2} < 0 \quad \text{for } x>0.
]
So (U) is concave for (x>0). In equilibrium choice problems, concavity ensures maxima rather than minima when optimizing.

1.5 Integration: Total Effects from Margins

Economic reasoning often uses integrals to recover totals from marginal relationships.

• If (m(x)) is a marginal effect, then total change from (a) to (b) is
  [
  \int_a^b m(x),dx.
  ]

Worked example: Total cost from marginal cost

Assume marginal cost
[
MC(q) = 2q + 3
]
and baseline cost at (q=0) is (C(0)=5).
Then
[
C(q)=C(0)+\int_0^q (2x+3),dx.
]
Compute integral:
[
\int_0^q (2x+3),dx = \left[x^2+3x\right]_0^q = q^2+3q.
]
Thus
[
C(q)=5+q^2+3q.
]
Interpretation: this is a typical way to reconstruct a cost curve from a marginal cost expression.

1.6 Series and Approximations (When Exact Answers Are Not Needed)

Series show up in approximations and in some econometrics derivations. Even if your exam focuses more on calculus, you may see Taylor expansions.

For a function (f(x)), the second-order Taylor approximation around (x=a) is:
[
f(x)\approx f(a) + f'(a)(x-a) + \frac{f''(a)}{2}(x-a)^2.
]
This is useful in:

• linearization (local approximations),
• interpreting elasticities near a point,
• simplifying nonlinear models.

Worked example: Linear approximation of (\ln(1+x)) around (x=0)

For small (x):
[
\ln(1+x)\approx x.
]
If (x=0.1), true value (\ln(1.1)\approx 0.0953), approximation (=0.1). In exam contexts, you may compare the approximation qualitatively (accuracy improves as (x\to 0)).

1.7 Exam Workflow for Calculus Questions

When calculus appears, examiners reward consistent structure. Use this checklist:

1. Rewrite the expression neatly.
2. Check domain (e.g., (\ln(x)) requires (x>0)).
3. Choose method (factorization vs. derivative vs. L’Hôpital only if needed).
4. Perform algebra carefully with signs and powers.
5. Interpret the result in economic language if it is an economics question (marginal effect, increasing/decreasing, equilibrium implication).

This process is particularly important in ECON613 because math is a language of economics; the math is not “just math.”

2) Optimization and Consumer/Producer Choice (Unconstrained and Constrained)

Optimization is the most frequent calculus-based topic in economics. ECON613 typically tests both:

• unconstrained optimization (set derivative(s) equal to zero),
• constrained optimization (budget constraints, technology constraints, Lagrangian).

This section also bridges to equilibrium interpretation and comparative statics.

2.1 Unconstrained Optimization: Finding Max/Min

To find maxima/minima of a function (f(x)):

1. Compute (f'(x)).
2. Solve (f'(x)=0) for critical points.
3. Use the second derivative (f''(x)) to classify:
   • If (f''(x)<0): local maximum.
   • If (f''(x)>0): local minimum.

Worked example: Profit maximization

Suppose a firm’s profit is:
[
\pi(q)=100q-2q^2-50.
]
Differentiate:
[
\pi'(q)=100-4q.
]
Set equal to zero:
[
100-4q=0 \Rightarrow q^=25.
]
Second derivative:
[
\pi''(q)=-4<0 \Rightarrow \text{maximum.}
]
Profit at (q^=25):
[
\pi(25)=100(25)-2(25)^2-50=2500-2(625)-50=2500-1250-50=1200.
]

In an exam, you should state both the optimal choice and the nature of extremum (maximum profit).

2.2 Constrained Optimization with Lagrange Multipliers

If you must maximize utility subject to a budget, or cost minimization subject to output requirements, you use the Lagrangian:
[
\mathcal{L}(x,y,\lambda)=\text{objective} + \lambda(\text{constraint}).
]

General steps:

1. Write the Lagrangian with multiplier (\lambda).
2. Take first-order conditions:
   [
   \frac{\partial \mathcal{L}}{\partial x}=0,\quad \frac{\partial \mathcal{L}}{\partial y}=0,\quad \frac{\partial \mathcal{L}}{\partial \lambda}=0.
   ]
3. Solve the system.
4. Check second-order conditions where appropriate (for concave objective and convex constraints, FOCs are typically sufficient).

2.3 Consumer Choice: Utility Maximization Under a Budget Constraint

Consider a consumer with utility:
[
U(x,y)=x^{0.5}y^{0.5}
]
subject to:
[
p_x x + p_y y \le m.
]
Typically, at optimum the budget binds (for monotonic utilities), so:
[
p_x x + p_y y = m.
]

Worked example: Solve the Cobb-Douglas case

Let (p_x=10), (p_y=20), and income (m=1000).

Lagrangian:
[
\mathcal{L}=x^{0.5}y^{0.5}+\lambda(1000-10x-20y).
]
FOCs:
[
\frac{\partial \mathcal{L}}{\partial x}=0.5x^{-0.5}y^{0.5}-10\lambda=0,
]
[
\frac{\partial \mathcal{L}}{\partial y}=0.5x^{0.5}y^{-0.5}-20\lambda=0,
]
[
\frac{\partial \mathcal{L}}{\partial \lambda}=1000-10x-20y=0.
]

Divide the first two FOCs to eliminate (\lambda):
[
\frac{0.5x^{-0.5}y^{0.5}}{0.5x^{0.5}y^{-0.5}} = \frac{10\lambda}{20\lambda}=\frac{1}{2}.
]
Left side simplifies:
[
\frac{x^{-0.5}y^{0.5}}{x^{0.5}y^{-0.5}} = x^{-1}y^{1}= \frac{y}{x}.
]
So:
[
\frac{y}{x}=\frac{1}{2} \Rightarrow y=\frac{x}{2}.
]

Plug into budget:
[
10x+20\left(\frac{x}{2}\right)=10x+10x=20x=1000 \Rightarrow x^=50.
]
Then:
[
y^=\frac{50}{2}=25.
]

Economic interpretation: With Cobb-Douglas (x^{0.5}y^{0.5}), each good gets fixed budget shares equal to exponents (here 50% each). Indeed:

• Expenditure on (x): (10x=10\cdot 50=500) (50% of 1000)
• Expenditure on (y): (20y=20\cdot 25=500)

In a marking scheme, showing the budget share logic can complement algebra, but you still must produce correct numerical solutions.

2.4 Duality and the Meaning of the Lagrange Multiplier

The Lagrange multiplier (\lambda) has an economic interpretation: under standard conditions, it equals the marginal utility of income.

Using the above problem, (\lambda) indicates how much utility increases when income increases marginally (holding prices constant). In many ECON613 exam solutions, you’re expected to compute (\lambda) or at least interpret it qualitatively.

Continuing with:
[
\frac{\partial \mathcal{L}}{\partial x}=0.5x^{-0.5}y^{0.5}-10\lambda=0
]
So:
[
10\lambda = 0.5 x^{-0.5} y^{0.5} = 0.5\sqrt{\frac{y}{x}}.
]
But (y/x=1/2), so:
[
\sqrt{\frac{y}{x}}=\sqrt{\frac{1}{2}}=\frac{1}{\sqrt{2}}.
]
Thus:
[
10\lambda = 0.5\cdot \frac{1}{\sqrt{2}} = \frac{0.5}{\sqrt{2}}=\frac{1}{2\sqrt{2}}.
]
So:
[
\lambda = \frac{1}{20\sqrt{2}}.
]
You may keep it in exact form; decimals are usually optional.

Interpretation: (\lambda) tells you the utility change per unit of income. If the exam asks for a sentence: “(\lambda) is marginal utility of income.”

2.5 Producer Choice: Cost Minimization and Supply

Producers typically:

• maximize profits by choosing inputs, or
• minimize cost subject to achieving output.

A common setup in intermediate micro/ECON613 is:
[
\min_{K,L} wL + rK \quad \text{s.t.}\quad q=f(K,L).
]
Where (w) is wage, (r) is rental rate of capital.

Worked example: Leontief production (perfect complements)

Let production be:
[
q = \min{K, 2L}.
]
To produce (q), you need:
[
K=q,\quad 2L=q \Rightarrow L=\frac{q}{2}.
]
Cost:
[
C(q)=wL + rK = w\left(\frac{q}{2}\right)+r(q) = q\left(\frac{w}{2}+r\right).
]
If (w=50), (r=100), then:
[
C(q)=q\left(\frac{50}{2}+100\right)=q(25+100)=125q.
]
Marginal cost:
[
MC(q)=C'(q)=125.
]
This example highlights: with Leontief technology, cost can be linear in (q), producing constant marginal cost.

Worked example: Cobb-Douglas production (more complex)

If:
[
q = K^{0.3}L^{0.7},
]
and you minimize cost with given output (q), you typically derive input demands:
[
K(q)=\alpha q,\quad L(q)= (1-\alpha)q
]
in a Cobb-Douglas log-linear way only under special parameterizations; otherwise the solution involves solving the Lagrangian and using FOCs.

In exam preparation, focus on the mechanics:

• Write Lagrangian
• Take FOCs w.r.t. (K) and (L)
• Use the constraint to solve
• Interpret the multiplier as the shadow cost of output

2.6 Comparative Statics: How Optima Change with Prices and Income

After finding optimal choices, econ exams frequently ask:

• How does (x^*) change when price (p_x) rises?
• How does equilibrium output change when costs shift?

In calculus-based comparative statics, you can use:

• implicit differentiation,
• sign arguments using derivatives and curvature.

Example: Cobb-Douglas demand comparative statics

In the earlier Cobb-Douglas problem (x^=0.5\cdot \frac{m}{p_x}).
With (x^ = \frac{1}{2}\frac{m}{p_x}), we can see:

• If (p_x) increases, (x^*) decreases linearly with (1/p_x).
• Elasticity of (x) w.r.t. (p_x) equals (-1) times the share? For Cobb-Douglas, each good’s demand is proportional to income and inversely proportional to its own price with unit elasticity: for (U=x^{a}y^{1-a}), demand (x^=a\frac{m}{p_x}). Hence:
  [
  \ln x^ = \ln a + \ln m – \ln p_x
  ]
  so elasticity of (x^*) wrt (p_x) is (-1).

If the exam asks for a sign: (dp_x>0 \Rightarrow dx^*<0). If it asks for elasticity: it’s typically (-1) for Cobb-Douglas shares with expenditure share (a).

2.7 Lagrangian Pitfalls to Avoid

Common mistakes in ECON613 optimization:

• Forgetting the constraint binding at optimum.
• Mixing up partial derivatives.
• Incorrect algebra when dividing FOCs (sign errors).
• Not checking curvature: maxima/minima classification.
• Not stating final choices clearly (e.g., giving only (x/y) ratio without numeric solutions).

A good exam answer explicitly includes:

1. FOCs,
2. elimination/division step,
3. substitution into constraint,
4. final optimal values,
5. brief economic interpretation.

3) Linear Algebra for Economists (Systems, Matrices, Eigenvalues)

Linear algebra appears throughout economics: solving systems of equations, studying transformations, analyzing equilibrium conditions, and understanding properties of matrices used in econometrics (covariance matrices, projections).

3.1 Vectors, Matrices, and Basic Operations

A vector in (\mathbb{R}^n) is written as:
[
\mathbf{x}=(x_1,x_2,\dots,x_n)^\top.
]
A matrix (A) is an array of numbers (A_{ij}).

Key operations:

• Matrix addition/subtraction (same dimensions)
• Scalar multiplication
• Matrix multiplication:
  [
  (AB){ik}=\sum_j A{ij}B_{jk}.
  ]
• Transpose: (A^\top)
• Identity matrix (I) with (IA=A)

Worked example: Solve a linear system via matrices

Solve:
[
\begin{cases}
x+2y=5\
3x-y=4
\end{cases}
]
Write as (A\mathbf{z}=\mathbf{b}):
[
A=\begin{pmatrix}1&2\3&-1\end{pmatrix},\quad
\mathbf{z}=\begin{pmatrix}x\y\end{pmatrix},\quad
\mathbf{b}=\begin{pmatrix}5\4\end{pmatrix}.
]

You can solve by elimination:
From first: (x=5-2y).
Sub into second:
[
3(5-2y)-y=4 \Rightarrow 15-6y-y=4 \Rightarrow 15-7y=4.
]
So:
[
-7y=-11 \Rightarrow y=\frac{11}{7}.
]
Then
[
x=5-2\cdot \frac{11}{7}=\frac{35}{7}-\frac{22}{7}=\frac{13}{7}.
]

In ECON613, you may be asked to show the steps or compute determinants/inverses.

3.2 Determinants and Invertibility

For a (2\times 2) matrix:
[
A=\begin{pmatrix}a&b\c&d\end{pmatrix},
\quad \det(A)=ad-bc.
]
If (\det(A)\neq 0), (A) is invertible.

Worked example: Invertibility check

Let
[
A=\begin{pmatrix}2&1\4&2\end{pmatrix}.
]
Then:
[
\det(A)=2\cdot 2 – 1\cdot 4 = 4-4=0.
]
So the system might have infinitely many solutions or none, depending on (\mathbf{b}).

3.3 Matrix Inverses (When Needed)

If:
[
A=\begin{pmatrix}a&b\c&d\end{pmatrix}
]
with (\det(A)\neq 0), then
[
A^{-1}=\frac{1}{ad-bc}\begin{pmatrix}d&-b\-c&a\end{pmatrix}.
]

Then solution is (\mathbf{z}=A^{-1}\mathbf{b}).

Worked example: Compute equilibrium prices from linear demand/supply

Suppose market equilibrium occurs when:
[
Q_d = 100 – 5P,\quad Q_s = 20 + 3P.
]
Set (Q_d=Q_s):
[
100-5P=20+3P.
]
Then:
[
80=8P \Rightarrow P=10.
]
You can also write these as linear equations and solve with matrices if there are multiple markets/variables.

Even when the test question is simple, being able to set up the matrix formulation is useful.

3.4 Eigenvalues and Eigenvectors (Stability and Dynamics)

Eigenvalues (\lambda) and eigenvectors (\mathbf{v}) satisfy:
[
A\mathbf{v}=\lambda\mathbf{v}.
]
To find eigenvalues, solve:
[
\det(A-\lambda I)=0.
]

In economics, eigenvalues show:

• stability in dynamic systems,
• behavior of repeated transformations,
• principal components in econometrics.

Worked example: Eigenvalues of a simple matrix

Let
[
A=\begin{pmatrix}2&0\0&3\end{pmatrix}.
]
Then eigenvalues are directly (2) and (3) with eigenvectors corresponding to axes.

For more general matrices, compute determinant.

Exam note: If the exam includes eigenvalues, expect either:

• a small matrix (2×2 or simple 3×3),
• interpretation in words (e.g., which direction grows fastest).

3.5 Positive Definiteness and Quadratic Forms

Many optimization and statistics results use quadratic forms:
[
q(\mathbf{x}) = \mathbf{x}^\top A\mathbf{x}.
]
If (A) is symmetric and positive definite, then (q(\mathbf{x})>0) for (\mathbf{x}\neq 0). This links to:

• Hessian matrix in second-order conditions for optimization,
• convexity of functions.

Worked example: Check convexity via Hessian

If an objective is:
[
f(\mathbf{x}) = \frac{1}{2}\mathbf{x}^\top A \mathbf{x},
]
then the Hessian is (A). If (A) is positive definite, the objective is strictly convex.

In an exam, a common method for (2\times 2) symmetric (A):

• check leading principal minors are positive:
  • (A_{11}>0),
  • (\det(A)>0).

3.6 Projection and Least Squares (Bridge to Econometrics)

Least squares uses projections onto column spaces of a matrix. If:
[
\mathbf{y}=X\beta+\varepsilon,
]
the OLS estimator is:
[
\hat{\beta}=(X^\top X)^{-1}X^\top \mathbf{y}
]
assuming (X^\top X) is invertible.

While ECON613 may not be a full econometrics course, the linear algebra of OLS is often included early.

Worked example: Simple OLS with one regressor

Let (y) and (x) be observed vectors with (X=[\mathbf{1}, x]). Then:
[
\hat{\beta}=(X^\top X)^{-1}X^\top y.
]
In an exam, if the dataset is small, you may compute the estimator numerically.

3.7 Matrix Calculus Basics (Derivatives of Vectors/Matrices)

ECON613 may ask you to compute derivatives like:

• (\frac{\partial}{\partial x}(x^\top A x)),
• (\nabla_x (Ax)),
• gradients in multivariable optimization.

A key identity (for symmetric (A)):
[
\frac{\partial}{\partial x}\left(x^\top A x\right)= (A+A^\top)x = 2Ax.
]
If (A) is symmetric, this becomes (2Ax).

In optimization problems with multivariate functions, gradients and Hessians determine maxima/minima through:

• first-order condition: (\nabla f(\mathbf{x})=0),
• second-order condition: Hessian negative definite for maxima.

4) Probability and Statistics for Economists (Uncertainty, Expectation, and Distributions)

Even when ECON613 is titled “Mathematics for Economists,” probability and statistics are often assessed because they support econometrics and decision-making under uncertainty.

4.1 Random Variables and Probability Rules

A random variable maps outcomes to numerical values. Discrete random variables have probability mass functions (pmf); continuous random variables have density functions (pdf).

Core rules:

• (0\le P(A)\le 1)
• (P(\Omega)=1)
• Complement: (P(A^c)=1-P(A))
• Addition for disjoint events: if (A\cap B=\varnothing), then (P(A\cup B)=P(A)+P(B))

Worked example: Basic conditional probability

Suppose (P(A)=0.3) and (P(B|A)=0.5). Then:
[
P(A\cap B)=P(B|A)P(A)=0.5\cdot 0.3=0.15.
]

4.2 Conditional Probability and Independence

Conditional probability:
[
P(A|B)=\frac{P(A\cap B)}{P(B)} \quad \text{if } P(B)>0.
]
Independence:
[
A \text{ and } B \text{ independent } \Leftrightarrow P(A\cap B)=P(A)P(B).
]
Equivalently:
[
P(A|B)=P(A) \quad \text{if } P(B)>0.
]

Worked example: Determine independence

Let (P(A)=0.4), (P(B)=0.5), and (P(A\cap B)=0.2).
Check:
[
P(A)P(B)=0.4\cdot 0.5=0.2.
]
Matches (P(A\cap B)), so independent.

4.3 Expectation and Variance (Central to Econometrics)

Expectation of discrete (X):
[
\mathbb{E}[X]=\sum_x x\cdot P(X=x).
]
Expectation of continuous (X):
[
\mathbb{E}[X]=\int_{-\infty}^\infty x f(x),dx.
]

Variance:
[
\mathrm{Var}(X)=\mathbb{E}[(X-\mathbb{E}[X])^2] = \mathbb{E}[X^2]-\left(\mathbb{E}[X]\right)^2.
]
Standard deviation is (\sqrt{\mathrm{Var}(X)}).

Worked example: expectation and variance of a discrete variable

Let (X) take values (1,2,3) with probabilities (0.2,0.5,0.3).
Then:
[
\mathbb{E}[X]=1(0.2)+2(0.5)+3(0.3)=0.2+1+0.9=2.1.
]
Next:
[
\mathbb{E}[X^2]=1^2(0.2)+2^2(0.5)+3^2(0.3)=0.2+4(0.5)+9(0.3)=0.2+2+2.7=4.9.
]
Variance:
[
\mathrm{Var}(X)=4.9-(2.1)^2=4.9-4.41=0.49.
]
So (\mathrm{SD}(X)=0.7).

4.4 Expectation of Functions, Linearity, and Covariance

A major property:
[
\mathbb{E}[aX+bY]=a\mathbb{E}[X]+b\mathbb{E}[Y].
]
Even if (X) and (Y) are dependent, this linearity holds.

Covariance:
[
\mathrm{Cov}(X,Y)=\mathbb{E}[(X-\mathbb{E}[X])(Y-\mathbb{E}[Y])].
]
If independent, (\mathrm{Cov}(X,Y)=0) (though the converse isn’t always true).

Correlation:
[
\rho_{XY}=\frac{\mathrm{Cov}(X,Y)}{\sigma_X\sigma_Y}.
]

4.5 Common Distributions in Economic Contexts

Bernoulli and Binomial

If (X\sim \text{Bernoulli}(p)), then:
[
P(X=1)=p,\quad P(X=0)=1-p,\quad \mathbb{E}[X]=p,\quad \mathrm{Var}(X)=p(1-p).
]
If (X\sim \text{Binomial}(n,p)):
[
\mathbb{E}[X]=np,\quad \mathrm{Var}(X)=np(1-p).
]

Poisson

If number of events in a time interval follows Poisson with rate (\lambda):
[
X\sim \text{Poisson}(\lambda),\quad \mathbb{E}[X]=\lambda,\quad \mathrm{Var}(X)=\lambda.
]

Normal

If (X\sim N(\mu,\sigma^2)):
[
\mathbb{E}[X]=\mu,\quad \mathrm{Var}(X)=\sigma^2.
]
Normal distribution is central to inference.

4.6 Transformations and the Method of Moments (Occasionally)

If (Y=aX+b), then:
[
\mathbb{E}[Y]=a\mathbb{E}[X]+b,\quad \mathrm{Var}(Y)=a^2\mathrm{Var}(X).
]
This allows quick transformation of variables.

Example: Using scaling for variance

If (X) has variance (4), and (Y=3X-2), then:
[
\mathrm{Var}(Y)=3^2\cdot 4=9\cdot 4=36.
]

4.7 Conditional Expectation (Expectations Under Information)

Conditional expectation:
[
\mathbb{E}[X|Y]
]
represents the expected value of (X) given the information contained in (Y).

A common identity:
[
\mathbb{E}[\mathbb{E}[X|Y]]=\mathbb{E}[X]
]
(tower property). In econometrics this underlies unbiasedness and iterated expectations.

4.8 Exam Practice: Probability Computations Under Time Pressure

Use a structured approach:

1. Identify whether the variable is discrete or continuous.
2. Write pmf/pdf clearly.
3. Compute expectation using definition.
4. For independence, use factorization and simplify.
5. For conditional probability, compute joint probability then divide by (P(\text{condition})).

In ECON613, most marks are earned by correct setup more than by complex computation.

5) Calculus + Linear Algebra Applied to Econometric Reasoning (Estimation, Optimization Links, and Solution Presentation)

This final section integrates themes: optimization via gradients/hessians, linear algebra via least squares, and probability via expectations. ECON613 often tests whether you understand how these parts fit together, not just isolated techniques.

5.1 The Optimization View of Least Squares

Consider the least squares objective:
[
Q(\beta)=\sum_{i=1}^n (y_i – x_i^\top \beta)^2.
]
In matrix form:
[
Q(\beta) = (\mathbf{y}-X\beta)^\top(\mathbf{y}-X\beta).
]
Expand:
[
Q(\beta)=\mathbf{y}^\top\mathbf{y} -2\beta^\top X^\top \mathbf{y} + \beta^\top X^\top X \beta.
]

To minimize, set gradient to zero. The derivative:
[
\nabla_\beta Q(\beta) = -2X^\top \mathbf{y} + 2X^\top X\beta.
]
Set to zero:
[
-2X^\top \mathbf{y} + 2X^\top X\beta =0
\Rightarrow X^\top X \beta = X^\top \mathbf{y}
\Rightarrow \hat{\beta}=(X^\top X)^{-1}X^\top \mathbf{y}.
]
This shows the direct optimization—linear algebra bridge.

In an exam, you may be asked to derive this result quickly. The key is:

• write the objective in quadratic form,
• compute gradient correctly,
• use invertibility assumptions.

5.2 Second-Order Conditions and Convexity in Estimation

Because (Q(\beta)) is quadratic, and (X^\top X) is positive semidefinite, the objective is convex. If (X) has full column rank, (X^\top X) is positive definite, so you get a unique minimizer.

Second derivative (Hessian):
[
\nabla^2_\beta Q(\beta)=2X^\top X.
]
If positive definite, you have strict convexity.

Exam expectation: Mention that least squares has a unique minimizer under full rank.

5.3 Expectation of Estimators and Unbiasedness

In a standard linear model:
[
\mathbf{y}=X\beta+\varepsilon,
]
with (\mathbb{E}[\varepsilon|X]=0), then:
[
\mathbb{E}[\hat{\beta}|X]=\beta.
]
Proof:
[
\hat{\beta}=(X^\top X)^{-1}X^\top \mathbf{y}
=(X^\top X)^{-1}X^\top (X\beta+\varepsilon)
=\beta + (X^\top X)^{-1}X^\top \varepsilon.
]
Taking conditional expectation:
[
\mathbb{E}[\hat{\beta}|X]=\beta+(X^\top X)^{-1}X^\top \mathbb{E}[\varepsilon|X]=\beta.
]

In ECON613, you may be asked to state conditions for unbiasedness rather than compute actual values.

5.4 Variance of OLS and the Role of Covariance Matrices

If:
[
\mathrm{Var}(\varepsilon|X)=\sigma^2 I,
]
then:
[
\mathrm{Var}(\hat{\beta}|X)=\sigma^2 (X^\top X)^{-1}.
]
This depends on:

• (\sigma^2): error variance,
• ((X^\top X)^{-1}): design matrix information.

In exams, you might be asked:

• what happens to variance if you increase sample size (generally variance decreases under regular conditions),
• how multicollinearity affects estimation (near-singular (X^\top X) inflates ((X^\top X)^{-1})).

5.5 Solving a Full Optimization-to-Estimation Micro-Problem

Scenario (exam-style)

Let there be one regressor with intercept (two parameters). Suppose:
[
X=
\begin{pmatrix}
1 & 0\
1 & 1\
1 & 2
\end{pmatrix},
\quad
\mathbf{y}=
\begin{pmatrix}
1\
3\
5
\end{pmatrix}.
]
Compute (\hat{\beta}) using:
[
\hat{\beta}=(X^\top X)^{-1}X^\top \mathbf{y}.
]

Step 1: Compute (X^\top X).
[
X^\top X=
\begin{pmatrix}
1&1&1\
0&1&2
\end{pmatrix}
\begin{pmatrix}
1&0\
1&1\
1&2
\end{pmatrix}

\begin{pmatrix}
1+1+1 & 0+1+2\
0+1+2 & 0^2+1^2+2^2
\end{pmatrix}

\begin{pmatrix}
3 & 3\
3 & 5
\end{pmatrix}.
]

Step 2: Compute (X^\top \mathbf{y}).
[
X^\top \mathbf{y}=
\begin{pmatrix}
1&1&1\
0&1&2
\end{pmatrix}
\begin{pmatrix}
1\
3\
5
\end{pmatrix}

\begin{pmatrix}
1+3+5\
0\cdot 1+1\cdot 3+2\cdot 5
\end{pmatrix}

\begin{pmatrix}
9\
13
\end{pmatrix}.
]

Step 3: Invert (X^\top X).
[
\det(X^\top X)=3\cdot 5 – 3\cdot 3=15-9=6.
]
[
(X^\top X)^{-1}=\frac{1}{6}\begin{pmatrix}
5 & -3\
-3 & 3
\end{pmatrix}.
]

Step 4: Multiply:
[
\hat{\beta}=(X^\top X)^{-1}X^\top \mathbf{y}
=\frac{1}{6}\begin{pmatrix}
5 & -3\
-3 & 3
\end{pmatrix}
\begin{pmatrix}
9\
13
\end{pmatrix}.
]
Compute vector:
[
\begin{pmatrix}
5\cdot 9 -3\cdot 13\
-3\cdot 9 +3\cdot 13
\end{pmatrix}

\begin{pmatrix}
45-39\
-27+39
\end{pmatrix}

\begin{pmatrix}
6\
12
\end{pmatrix}.
]
Then:
[
\hat{\beta}=\frac{1}{6}\begin{pmatrix}6\12\end{pmatrix}

\begin{pmatrix}
1\
2
\end{pmatrix}.
]

Interpretation: the fitted model is ( \hat{y}=1+2x), which matches the observed points exactly: at (x=0), (y=1); at (x=1), (y=3); at (x=2), (y=5).

This micro-problem demonstrates a complete chain: matrix setup → compute normal equations → invert → produce coefficients.

5.6 Presenting Solutions Like an Exam Winner

ECON613 marking often follows a “process + correctness” style. Your answers should include:

• correct formulas,
• clear substitution,
• structured algebra,
• final numerical/closed-form results,
• short interpretation.

A good “presentation template” for optimization problems:

1. Define objective and constraint(s).
2. Write Lagrangian (if constrained).
3. Compute FOCs.
4. Solve system step-by-step.
5. State optimum and classify (max/min) if asked.
6. Provide economic meaning (e.g., marginal utility of income, shadow price, comparative static sign).

A good “presentation template” for least squares:

1. Write objective (Q(\beta)).
2. Expand into quadratic form.
3. Compute gradient and set to zero.
4. Derive normal equations.
5. Solve for (\hat{\beta}) using matrix inverse (with determinant check for invertibility).
6. Provide fitted equation and a one-line interpretation.

5.7 South African Context: Practical Study Strategy for ECON613

While exam formats vary across South African universities and TVET colleges, the mathematical rigor tends to follow the same core set: calculus for optimization, linear algebra for systems/OLS, and probability for modeling uncertainty. A strategy that consistently performs well includes:

• Daily “micro-drills” (30–45 minutes):
  • 10 minutes: differentiation/integration practice,
  • 10 minutes: optimization steps (FOCs + second derivative),
  • 10–15 minutes: one linear algebra operation (inverse, determinant, eigenvalues),
  • remainder: a probability computation (expectation/variance).
• Weekly “full problem sets”:
  • one long optimization problem with Lagrangian,
  • one linear algebra system/OLS derivation,
  • one probability question combining conditional expectation and variance.
• Error log:
  • track where mistakes occur: sign errors, wrong derivative, incorrect substitution, wrong interpretation.
• Interpretation habit:
  • after every mathematical answer, write 1 sentence linking it to economic meaning.

Because ECON613 is mathematics for economists, an answer that is mathematically correct but economically unexplained can still lose marks when interpretation is explicitly required.

5.8 High-Frequency Exam Topics Checklist (Use for Revision)

Before an ECON613 exam, ensure you can do the following without hesitation:

1. Differentiate and interpret marginal effects ((f')) and curvature ((f'')).
2. Solve unconstrained optimization (critical points + classification).
3. Set up and solve Lagrangian constrained problems (budget, technology, cost minimization).
4. Perform basic comparative statics using algebra and sign reasoning.
5. Solve linear systems using elimination and matrix inverses.
6. Use determinants to check invertibility.
7. Compute eigenvalues for simple matrices (2×2 focus).
8. Recognize positive definiteness and connect it to convexity/second-order conditions.
9. Derive OLS estimator from minimizing a quadratic objective.
10. Use expectation/variance properties and conditional probability rules.

5.9 Final Integrated Example (One Problem, Many Skills)

Suppose an economist models a simplified “policy choice” (x) that maximizes a net benefit:
[
\max_{x\ge 0} ; B(x)= -x^2 + 6x.
]

1. Find the optimum:
   [
   B'(x)=-2x+6=0 \Rightarrow x^*=3.
   ]
2. Check classification:
   [
   B''(x)=-2<0 \Rightarrow \text{maximum.}
   ]
   Net benefit:
   [
   B(3)=-(9)+18=9.
   ]

Now suppose uncertainty: (x) influences expected outcomes but policy effectiveness has error:
[
Y = B(x) + \varepsilon
]
with (\mathbb{E}[\varepsilon]=0) and (\mathrm{Var}(\varepsilon)=4).
Then:

• Expected outcome at (x^):
  [
  \mathbb{E}[Y|x^] = B(3) + \mathbb{E}[\varepsilon]=9.
  ]
• Variance of outcomes:
  [
  \mathrm{Var}(Y|x^*)=\mathrm{Var}(\varepsilon)=4
  ]
  (since (\varepsilon) variance is constant and independent of choice in this simplified model).

This combined problem shows what ECON613 demands: optimize deterministically, then interpret expectation/variance under uncertainty. In real exams, the “policy variable” could be an input choice, consumption, a coefficient in regression, or a decision under constraints; but the logic is the same.

6) Targeted Exam Templates for ECON613 (Institution-Independent, Institution-Centered Practice)

Different South African institutions may label course outcomes slightly differently, but the math tasks are consistent. This section provides institution-independent templates but organized as cluster-style study routines that resemble how many students structure modules and week-by-week preparation in universities and TVET colleges.

6.1 Cluster Routine A: “Calculus to Optimization” (60–90 minute block)

Use this routine when your exam includes Lagrange multipliers or second-order tests.

Step-by-step plan

1. Choose one unconstrained problem.
2. Do:
   • derivative,
   • set to zero,
   • second derivative sign,
   • compute optimum value.
3. Choose one constrained problem.
4. Do:
   • Lagrangian setup,
   • FOCs,
   • solve system,
   • interpret multiplier.
5. End with a comparative statics question:
   • vary a parameter symbolically or plug numbers to check direction.

Example set (template)

• Unconstrained: maximize (f(x)=ax – bx^2).
• Constrained: maximize (U(x,y)=x^{0.5}y^{0.5}) subject to (p_x x+p_y y=m).
• Comparative statics: increase (p_x) and compute new (x^*).

You already computed one full Cobb-Douglas solution earlier; re-do it under a different numerical set:

• keep the structure the same,
• focus on avoiding algebra slips.

6.2 Cluster Routine B: “Linear Systems + OLS” (60–90 minute block)

This routine matches exam questions that ask you to “show how you arrive at (\hat{\beta})” or solve equilibrium systems.

Step-by-step plan

1. Solve a system using elimination and then verify via matrix inversion.
2. Compute a small (X^\top X), its determinant, and inverse.
3. Derive (\hat{\beta}) using ((X^\top X)^{-1}X^\top y).
4. Provide a short fitted equation interpretation.

Micro-drill

• Given a 3-point dataset, build (X) with intercept and slope.
• Compute (\hat{\beta}) in full steps.
• If results match exactly, explain why (perfect linear relationship in data).

6.3 Cluster Routine C: “Probability + Expected Value for Economic Decisions” (45–75 minute block)

This routine supports econometrics-style reasoning questions about uncertainty.

Step-by-step plan

1. Compute conditional probabilities from joints.
2. Compute expectation and variance from definitions.
3. Use linearity of expectation to combine terms.
4. Use transformations for scaling and shifting.
5. Finish with an interpretation sentence (“expected cost,” “risk,” “mean-zero error implies unbiasedness”).

Example drill

• Let (X\sim\text{Binomial}(n,p)). Compute (\mathbb{E}[X]) and (\mathrm{Var}(X)).
• Let (Y=3X-2). Compute (\mathbb{E}[Y]), (\mathrm{Var}(Y)).

6.4 Cluster Routine D: “Integration of Skills” (90–120 minute block)

Some ECON613 questions combine methods. Prepare by practicing integrated problems.

Integrated problem template

1. Start with optimization to find a decision variable (x^*).
2. Substitute (x^*) into an expected value or profit expression.
3. Add uncertainty with an error term ( \varepsilon ) and compute:
   • expectation,
   • variance,
   • (if required) standard deviation.

This mirrors the typical econometrics-to-theory flow:

• derive the decision rule,
• compute expected outcomes,
• discuss variability.

6.5 Final Self-Assessment Checklist (Score Yourself Honestly)

Before exam week, verify you can answer quickly:

• Can you differentiate and simplify under chain rule without errors?
• Can you solve a Cobb-Douglas Lagrangian problem and produce correct numeric demands?
• Can you set up and solve (\hat{\beta}=(X^\top X)^{-1}X^\top y) for a small matrix?
• Can you compute (\mathbb{E}[X]) and (\mathrm{Var}(X)) for a three-outcome discrete variable?
• Can you explain curvature (concavity/convexity) in one sentence?

If you fail any item, target it with daily micro-drills until consistent.

6.6 Consistent Notes Format (How to Write High-Mark Answers)

Adopt a consistent solution writing format to avoid losing marks for “presentation gaps”:

• Underline the objective in optimization problems.
• Box the final answer(s).
• In Lagrangian problems, list FOCs explicitly.
• In matrix problems, write intermediate matrices ((X^\top X), (X^\top y), determinant).
• In probability problems, label events and show substitution steps.

Consistency signals clarity to markers and reduces accidental omissions.

6.7 Summary of What ECON613 Math Is Really Testing

ECON613 Math for Economists tests whether you can:

1. Model economic relationships as mathematical functions,
2. Optimize choices and interpret marginal effects,
3. Solve systems using linear algebra,
4. Reason under uncertainty using probability and expectation,
5. Connect calculus and linear algebra to econometric estimation.

A strong performance comes from fluency plus explanation—getting the numbers right and making the economic meaning explicit.