ECON31031 Cheatsheet · Advanced Econometrics

The University of Manchester · handwritten A4 planner for the final exam. Two-hour on-campus invigilated examination in January 2027 (check the date in the university exam schedule). Full ECON31031 study guide →

A4 landscape · 7pt · six columns per side · copy side 1 by handECON31031 · two sides
ECON31031
Advanced Econometrics
The University of Manchester · Level 3 · Semester 1, 2026/27
ECON31031 cheatsheet QR code – open the full ECON31031 study guide
Full guide + ask Sia
about ECON31031
SIDE 1/2   HANDWRITTEN A4 PLANNER · priority formulas · decision rules · exam answer scaffoldscopy this side by handFinal 75% · January 2027

0 · Exam mapcourse.json · brief

Coursework 25%Final Exam 75%

“Students will be allowed to bring one hand written A4 sheet of notes into the exam room.” (ECON31031 course page, Canvas, 2026/27 1st semester)

Printed sheets are not allowed. Use side 1 as the plan for the permitted handwritten page; use both sides of your handwritten A4.

A · microdata: answer 1 of 2 B · macrodata: answer 2 of 3
Paper choice: one Section A microdata answer, two Section B macrodata answers.
Put on the handwritten pageWhy
1. Time seriestwo of three answers
2. MLE + testsshared machinery
3. Binary calcsreliable Section A work
4. Matrix OLSderivations + testing
Sia → For every answer: set up → calculate/derive → interpret. State the distribution, degrees of freedom and conclusion for every test.

0b · Scientific-calculator workflowstudy strategy

  1. Write the model and assumptions before arithmetic.
  2. Keep full precision; round only the reported result.
  3. Mark known values, forecasts and unobserved shocks distinctly.
  4. End with one sentence answering the economic/statistical question.

Three answer skeletons

Derive: define the model/density and dimensions → state the assumptions → show the algebra using a named identity → box the estimator/result → say why it holds. Calculate: write the formula → substitute visible numbers → keep the denominator and units explicit → report a rounded value → interpret. Test: H₀/H₁ → statistic → reference distribution and df → critical value/decision → conclusion in the language of the question.

What earns space

KeepCompress
conditions changing the methodintroductory prose
denominators + dfsoftware commands
sign/scale interpretationrepeated definitions
one numerical check per familyderivations you can rebuild

1 · Matrix OLS corech.1

model + estimatory = Xβ + ε
β̂ = (X′X)⁻¹X′y

With n observations and k regressors plus intercept: y,ε are n×1; X is n×(k+1); β is (k+1)×1. Gauss–Markov gives BLUE under correct linear specification, E(ε|X)=0, no perfect collinearity and homoscedastic uncorrelated errors.

mean, variance, precisionE(β̂|X)=β
Var(β̂|X)=σ²(X′X)⁻¹
s²=e′e/(n−k−1)

yŷ=Pye=Mycol(X)
OLS projects y onto col(X); the residual My is orthogonal to every X column.

projection identitiesP=X(X′X)⁻¹X′; M=I−P
ŷ=Py; e=My=Mε; X′e=0
P′=P, P²=P; M′=M, M²=M; MX=0
tr(M)=n−k−1

Where it bites: derive unbiasedness/variance, prove s² unbiased, or use X′e=0 to catch hand-arithmetic errors.

1b · Fit and restrictionsch.1 FAQ

R²=1−RSS/TSS
F=[(RSSᵣ−RSSᵤ)/q] ÷ [RSSᵤ/(n−k−1)]

QuestionStatisticReference
one β restrictiont=(β̂ⱼ−βⱼ⁰)/setₙ₋ₖ₋₁ if normal
q joint restrictionsF aboveFq,n−k−1 if normal
large sampleqFχ²q

Write restrictions as Rβ=r. Ordinary R² never falls when regressors enter; adjusted R² can fall because it accounts for degrees of freedom.

Gauss–Markov line-by-line

Use a random cross-section and a model linear in β; require full column rank, E(ε|X)=0 and Var(ε|X)=σ²I. Then β̂ is unbiased and BLUE. Normal errors are an extra condition for exact finite-sample t/F inference; without normality, invoke large-sample normal/χ² results.

Restriction workflow

  1. Translate words into each row of Rβ=r.
  2. Estimate unrestricted model and record RSSᵤ.
  3. Impose/substitute restrictions; estimate and record RSSᵣ.
  4. Use q rows of R in F; RSSᵣ cannot be below RSSᵤ.

2 · Maximum likelihood recipech.2

  1. State one-observation density and support.
  2. Use independence: L(θ)=∏fᵢ(θ).
  3. Log: ℓ(θ)=Σ ln fᵢ(θ).
  4. Score: s(θ)=∂ℓ/∂θ; solve s(θ̂)=0.
  5. Curvature: H(θ̂)=∂²ℓ/∂θ∂θ′.
  6. Test restrictions and interpret.

large-sample covarianceVar̂(θ̂)=[−H(θ̂)]⁻¹

p̂=.301.92 below max.13.30.52
Six successes in twenty: ℓ peaks at .30; the LR acceptance band is .13–.52.

likelihood-ratio testLR=2[ℓ(θ̂ᵤ)−ℓ(θ̂ᵣ)] ~ χ²q
q = number of restrictions

Count restrictions, not parameters. A closed form exists only when the score equations solve algebraically; nonlinear index models normally require numerical optimisation.

2b · Two benchmark likelihoodsch.2

Bernoulliℓ(p)=Σ[yᵢ ln p+(1−yᵢ)ln(1−p)]
p̂=ȳ; Var̂(p̂)=p̂(1−p̂)/n

normal linear modelβ̂ML=β̂OLS
σ̂²ML=RSS/n; s²OLS=RSS/(n−k−1)

Both variance estimators are consistent; only the OLS version is unbiased in finite samples.

What each derivative supplies

ObjectUseExam line
L(θ)sample probability/densityproduct by IID
ℓ(θ)workable objectivesum of log densities
scorefirst-order conditionsset every component to zero
Hessiancurvature/precisioninvert its negative
ℓᵤ,ℓᵣjoint testLR=2(ℓᵤ−ℓᵣ)

For binary data, the intercept-only restricted model has fitted probability equal to the sample proportion. This gives ℓᵣ directly and reappears in likelihood-based pseudo-R² measures. A sharp peak means larger negative curvature and a smaller covariance; a flat likelihood means imprecision.

3 · Binary model chooserch.3

Modelpᵢ=Pr(yᵢ=1|xᵢ)Key exam point
LPMxᵢ′βeasy OLS; invalid p possible; constant effects; robust SE
LogitΛ(z)=eᶻ/(1+eᶻ)log[p/(1−p)]=z; OR=eᵝ
ProbitΦ(z)normal table; latent threshold
logitprobitLPM1.50
LPM is linear and can leave [0,1]; logit/probit approach the bounds smoothly.

LPM: Var(εᵢ|xᵢ)=pᵢ(1−pᵢ), so heteroscedasticity is built in. White robust covariance repairs inference, not out-of-range predictions or constant effects.

latent probityᵢ*=xᵢ′β+εᵢ; yᵢ=1[yᵢ*>0]
εᵢ~N(0,1) ⇒ Pr(yᵢ=1)=Φ(xᵢ′β)

Only the threshold side is observed, so error variance must be fixed. This scale normalisation explains why logit and probit coefficients differ even when fitted probabilities are close.

3b · Probability, odds, interpretationch.3 FAQ

odds=p/(1−p)=eᶻ; p=odds/(1+odds)
one-unit xⱼ rise ⇒ odds × eᵝʲ

For a dummy, compare odds at xⱼ=1 and 0 holding other regressors fixed. Never multiply the probability by eᵝ.

GLM and response type

The generalised-linear view changes the link: identity for LPM, logit for logistic probabilities, inverse-normal for probit. Binary outcomes have two unordered values; ordered responses rank categories without equal spacing; multinomial responses choose among several exclusive alternatives.

Limited dependent variables

Observed ruleModel idea
only which side of thresholdbinary response
threshold recorded below limitcensoring
observations below limit absenttruncation

Probit by hand: calculate z=x′β, then read Φ(z). For z<0 use symmetry, Φ(−a)=1−Φ(a). Scale warning: logistic error variance is π²/3 while standard normal variance is 1; coefficients are not directly comparable. In the course example, logit coefficients are roughly 1.6–1.8 times probit coefficients, while probability effects are close.

4 · Marginal effectsch.4

continuous regressorLPM: MEⱼ=β̂ⱼ
Logit: MEⱼ=p̂(1−p̂)β̂ⱼ
Probit: MEⱼ=φ(x′β̂)β̂ⱼ

dummy regressorΔp=F(x′β̂; xⱼ=1)−F(x′β̂; xⱼ=0)

The density is positive, so effect sign matches coefficient sign. For logit, p(1−p) peaks at .25 when p=.5; effects shrink near 0 or 1. Report probability changes in percentage points and scale them to the change asked.

SummaryCalculationSay
effect at averageME at x̄effect for mean profile
average effectn⁻¹ΣMEᵢmean individual effect
dummyexact p(1)−p(0)not derivative approximation

Use the delta method to transform Var(β̂) into an effect’s standard error. Odds ratios are logit-specific and constant across individuals; probability changes depend on the starting index.

5 · Binary-model assessmentch.5

At cut-off c: predict 1 when p̂≥c. In the hit/miss table let TP=n₁₁, FP=n₁₀, FN=n₀₁, TN=n₀₀.

MeasureFormula
hit count(TP+TN)/n
recall / TPRTP/(TP+FN)
FPRFP/(FP+TN)
KuipersTPR−FPR
precisionTP/(TP+FP)
F12·precision·recall/(precision+recall)

Brier=n⁻¹Σ(yᵢ−p̂ᵢ)²  [lower better]
McFadden R²=1−ℓᵤ/ℓᵣ

A high hit count misleads with imbalance. Prefer precision/recall/F1 and the ROC: plot TPR against FPR across cut-offs; AUC uses trapezia. Guide bands: .5–.7 poor; .7–.8 fair; .8–.9 good; >.9 excellent.

larger AUCchanceTPRFPR
ROC varies the cut-off: move towards top-left for high TPR and low FPR.

phi coefficientφ=(TP·TN−FP·FN)/√[(TP+FP)(FN+TN)(TP+FN)(FP+TN)]

In-sample fitted values describe the estimation sample; out-of-sample predictions evaluate observations not used in estimation. Do not call in-sample fit evidence of forecast performance.

6 · Stationarity + operatorsch.6

Lyₜ=yₜ₋₁; Lʰyₜ=yₜ₋ₕ
Δyₜ=(1−L)yₜ=yₜ−yₜ₋₁
seasonal s: Δₛyₜ=(1−Lˢ)yₜ

Weak stationarity: E(yₜ)=μ finite and constant; Var(yₜ)=γ₀ finite and constant; Cov(yₜ,yₜ₋ₖ)=γₖ depends only on lag k. White noise has mean 0, constant variance σ² and zero autocorrelation for k≠0; it need not be normal.

sample autocorrelationrₖ=Σₜ₌ₖ₊₁ᵀ(yₜ−ȳ)(yₜ₋ₖ−ȳ) / Σₜ₌₁ᵀ(yₜ−ȳ)²

Same denominator at every lag. Slow ACF decay warns of strong persistence or nonstationarity; do not apply stationary formulas until diagnosed.

7 · AR/MA/ARMA identificationch.7

AR(1)ACF decaysPACF cuts MA(1)ACF cutsPACF decays
Eyeball rule: AR cuts in the PACF; MA cuts in the ACF; mixed decay suggests ARMA.

AR(1): yₜ=α+φyₜ₋₁+εₜ|φ|<1; μ=α/(1−φ)
Var(y)=σ²/(1−φ²); ρₖ=φᵏ

MA(q): yₜ=α+εₜ+Σθⱼεₜ₋ⱼE(y)=α; Var(y)=σ²(1+Σθⱼ²)
MA(1): ρ₁=θ/(1+θ²); ρₖ=0 for k>1

AR(2) stationarity: φ₁+φ₂<1; φ₂−φ₁<1; φ₂>−1. Equivalently check all characteristic roots have absolute value below one. MA terms need invertibility.

8 · Forecast, select, differencech.8–9

AR(1): ŷT+h|T=μ+φʰ(yT−μ)
MA(1): h=1 uses θêT; h≥2 → μ
AIC=−2ℓ+2K; BIC=−2ℓ+K ln T

Feed forecasts back when extrapolating; if actual out-of-sample values arrive, use them. Minimise AIC/BIC; BIC penalises dimension more heavily once T exceeds about seven.

random walk: tσ²AR(.8) → 2.78Var
With unit shock variance, stationary AR(.8) settles at 2.78; random-walk variance grows.
https://www.asksia.ai/cheatsheet/manchester-econ31031 · Full ECON31031 study guide →
Forward this to your tutorial group · Shared by ________
AskSiav2 · 2026-10-19 · side 1/2
Two-hour on-campus invigilated examination in January 2027 (check the date in the university exam schedule).
Your exam allows one HANDWRITTEN A4 (both sides): copy side 1 by hand; do not bring this printout.
ECON31031
Advanced Econometrics · Worked Drills
five source-grounded examples · traps · fast recall
ECON31031 cheatsheet QR code – open the full ECON31031 study guide
Full guide + ask Sia
about ECON31031
SIDE 2/2   DRILLS · complete calculations · common traps · method chooserpractise before the examderive · calculate · interpret

W1 · Matrix OLS by handch.1 worked example

Data: (x,y)=(0,1),(1,3),(2,4),(3,6). Estimate intercept/slope, s² and test H₀:β₁=0.

1 · cross-productsX′X=[[4,6],[6,14]], det=20
(X′X)⁻¹=(1/20)[[14,−6],[−6,4]]
X′y=(14,29)′

2 · estimates + residualsβ̂=(22/20,32/20)′=(1.1,1.6)′
ŷ=(1.1,2.7,4.3,5.9)′
e=(−.1,.3,−.3,.1)′

3 · uncertainty + teste′e=.20; df=4−1−1=2; s²=.10
Var̂(β̂₁)=.10(4/20)=.02
se=.141; t=1.6/.141=11.3

Result: ŷ=1.1+1.6x; s²=.1; t=11.3 on 2 df rejects a zero slope.

yPyMycol(X)
Hand-check geometry: fitted vector lies in col(X); residual is perpendicular.
Check → Σe=0 and X′e=0 before using residuals. These catch a wrong inverse or multiplication.

T1 · OLS trapsch.1 FAQ

TrapCorrection
divide RSS by nunbiased s² uses n−k−1
forget q in joint testF numerator divides by q; large sample qF~χ²q
restriction unclearwrite Rβ=r before algebra
add regressor ⇒ fit “improves”R² cannot fall; inspect adjusted R²

Restriction self-check

Four coefficients β₁ (intercept) to β₄; impose β₂=2β₃ and β₁+β₄=0:

R=[[0,1,−2,0],[1,0,0,1]]; r=(0,0)′
β₂=2β₃; β₄=−β₁
y=β₁(1−x₄)+β₃(2x₂+x₃)+ε

Estimate the transformed equation with two constructed regressors. Two independent rows of R mean q=2. If restrictions already hold in the sample, imposing them leaves estimates/RSS unchanged and F=0.

Residual-maker derivation drill

e=My=M(Xβ+ε)=Mε because MX=0
E(e′e|X)=E[tr(Mεε′)|X]
=tr[M·σ²I]=σ²tr(M)

Two-by-two inverse card

Test answer checklist

W2 · Bernoulli MLE + LRch.2 worked example

Data: 20 successes in 80 independent trials. Estimate p; test H₀:p=.5 at 5%.

1 · log-likelihood + scoreℓ(p)=20 ln p+60 ln(1−p)
s(p)=20/p−60/(1−p)=0
p̂=20/80=.25

2 · two maximaℓᵤ=20 ln .25+60 ln .75=−44.99
ℓᵣ=80 ln .5=−55.45

3 · LRLR=2(−44.99+55.45)=20.93
χ²₁,.95=3.841 ⇒ reject H₀

p̂=.25H₀:.52× vertical gap = LR
LR doubles the log-likelihood gap between unrestricted p̂ and restricted p=.5.

Verdict sentence: “LR=20.93 exceeds 3.841, so reject p=.5 at 5%.”

T2 · Likelihood trapsch.2

  • Write the one-observation density first; support errors poison every later line.
  • LR is 2(ℓᵤ−ℓᵣ), never reverse the signs.
  • Use q restrictions for χ² df.
  • At a maximum H is negative; covariance uses −H inverse.
  • ML normal variance uses n; OLS unbiased variance uses n−k−1.

Bernoulli curvature check

s(p)=Σ[yᵢ/p−(1−yᵢ)/(1−p)]
H(p)=−Σ[yᵢ/p²+(1−yᵢ)/(1−p)²]<0
se(p̂)=√[p̂(1−p̂)/n]

For W2, se=√[.25(.75)/80]=.0484. The negative Hessian confirms a maximum; the inverse curvature supplies the same large-sample variance.

LR answer frame

  1. Fit/unpack the unrestricted model and record ℓᵤ.
  2. Impose H₀, re-maximise and record ℓᵣ.
  3. Compute 2(ℓᵤ−ℓᵣ); it must be non-negative.
  4. Use χ²q, where q counts imposed restrictions.
  5. Compare, decide and interpret the restrictions jointly.

Closed form vs numerical: Bernoulli p̂=ȳ and normal-regression β̂ have algebraic solutions. Logit, probit and ARMA parameters generally sit inside nonlinear sums, so say “solve numerically”; do not pretend the score isolates β.

W3 · Logit probability + oddscourse worked example

Model: z=−.8+.06(study hours)+.7(course). At 15 hours, compare course=1 vs 0.

1 · indicesz₁=−.8+.06(15)+.7=.8
z₀=−.8+.06(15)=.1

2 · probabilitiesp₁=e⁰·⁸/(1+e⁰·⁸)=2.2255/3.2255=.690
p₀=e⁰·¹/(1+e⁰·¹)=1.1052/2.1052=.525
Δp=.165 = 16.5 percentage points

3 · odds ratioodds₁=2.2255; odds₀=1.1052
OR=2.2255/1.1052=2.014=e·⁷

The course doubles the odds for every profile, but its probability effect depends on the starting index.

slope .250 at z=0slope .105 at z=2
Same coefficient, different probability effect: the logistic density shrinks in the tails.

D3 · Probit effects drillch.4 worked example

At z=0, βcontinuous=.25, βdummy=.5, φ(0)=.3989, Φ(.5)=.6915:

continuous ME=.3989(.25)=.0997
exact dummy=Φ(.5)−Φ(0)=.1915
derivative approx=.3989(.5)=.1995

Use the exact probability difference for a dummy. Here the derivative slightly overstates it because the curve flattens from 0 to .5.

Odds-update self-check

Starting logit probability .8 gives odds .8/.2=4. A dummy coefficient .5 multiplies odds by e·⁵=1.6487, giving 6.595; convert back: 6.595/7.595=.868. The probability rises 6.8 points, not by 64.87%.

Effect reporting grid

InputActionInterpretation
continuous xdensity × βp change per unit x
dummy xF(z₁)−F(z₀)percentage-point switch
logit βeᵝodds multiplier
change ΔxME × Δxscale to question

Scale check: a probit effect −.065 per unit becomes −.0065 for a .1 rise—about two-thirds of a percentage point. Keep “probability”, “percentage points” and “odds” distinct.

Model traps

  • Robust LPM SEs repair heteroscedastic inference, not invalid fitted probabilities.

W4 · Classification + AUCch.5 worked material

At cut-off .5, the guide’s twelve-person table has TP=5, FP=2, FN=1, TN=4.

observed10 pred 1pred 0 TP 5FP 2FN 1TN 4 TPR=.833FPR=.333Prec=.714F1=.769
Confusion-matrix denominators: TPR uses actual 1s; precision uses predicted 1s.

hit=(5+4)/12=.750
Kuipers=5/6−2/6=.500
precision=5/7=.714; recall=5/6=.833
F1=2(5)/(10+2+1)=.769

AUC drill points: (0,0),(.1,.5),(.3,.8),(.6,.95),(1,1).

.1(.25)+.2(.65)+.3(.875)+.4(.975)
=.025+.130+.2625+.390=.8075 ⇒ good

AUC=.8075false positive rateTPR
Sort by FPR, then sum width × average endpoint height for each trapezium.

T4 · Assessment trapsch.5 FAQ

  • One cut-off is one ROC point, not the whole curve.
  • Lower Brier is better; applying it to 0/1 predictions gives 1−hit count.
  • With rare successes, accuracy can look high while separation is poor.
  • Link test: add η̂² and η̂³; significant powers signal misspecification but do not identify its cause.

Phi and denominator audit

φ=(5·4−2·1)/√[(7)(5)(6)(6)]
=18/√1260=.507

MeasureCondition onQuestion answered
recallactual successhow many successes found?
precisionpredicted successhow many alerts correct?
FPRactual failurehow many false alarms?
KuipersTPR−FPRskill beyond false alarms?

Threshold direction: raising the cut-off usually predicts fewer successes: recall/TPR falls, FPR falls and precision often rises. The ROC traces this trade-off without committing to one cut-off.

W5 · Time-series chainch.6–9 worked examples

A · sample autocorrelation

y=(2,4,3,5,1), ȳ=3, deviations=(−1,1,0,2,−2); denominator=10. Neighbour products sum −5:

Δy=(2,−1,2,−4); r₁=−5/10=−.5

B · AR(1) and MA(1) moments

AR: yₜ=3+.5yₜ₋₁+εₜ, Var(ε)=3. MA coefficient θ=−.4.

μAR=3/(1−.5)=6
VarAR=3/(1−.25)=4
ρ₂=.5²=.25
ρ₁,MA=−.4/(1+.16)=−.345; ρk>1=0

φ=.8: same sign, geometric decay φ=−.8: alternating, geometric decay
AR(1) ACF is φᵏ: coefficient sign determines smooth persistence or oscillation.

C · AR forecast recursion

yₜ=1+.8yₜ₋₁+εₜ, last y=10:

ŷ₁=1+.8(10)=9.0
ŷ₂=1+.8(9)=8.2
ŷ₃=1+.8(8.2)=7.56
μ=1/(1−.8)=5

When extrapolating, forecast feeds forecast. Never use an unobserved value as though it were data.

D · identification and stationarity checks

CandidateACFPACF
AR(p)decayscuts after p
MA(q)cuts after qdecays
ARMA(p,q)decaysdecays
unit-root warningvery slow decaydo not force stationary rule

MA logic: autocovariances vanish beyond q because no shocks are shared. Invertibility rewrites the shock polynomial as an autoregression in observed y.

E · operator micro-drills

trend: yₜ=2t+εₜ ⇒ not stationary since E(yₜ)=2t
Δyₜ=2+εₜ−εₜ₋₁ ⇒ E=2, Var=2σ²
quarterly annual change: (1−L⁴)yₜ=yₜ−yₜ₋₄

Stationarity answers must name the failing condition. A plot or slow correlogram is evidence to investigate; a time-dependent mean/variance or a unit root is the formal demonstration.

F · MA and mixed forecasts

W5 continued · selection + unit rootch.8–9

base y=97.406.44mean 5
Source forecast path: α=2, φ=.6, yT=9; forecasts converge from 7.40 to 5.

D · information criteria

T=100; maximised ℓ: ARMA(1,0) −142.3; (2,0) −140.1; (1,1) −139.0; (2,2) −136.9. Count K=p+q+1.

ModelKAICBIC
(1,0)2288.6293.81
(2,0)3286.2294.02
(1,1)3284.0291.82
(2,2)5283.8296.83

AIC chooses (2,2); BIC chooses (1,1). Circle the minimum and explain BIC’s heavier penalty.

E · random walk with drift

yₜ=.5+yₜ₋₁+εₜ; y₀=10; Var(ε)=2. At t=8:

y₈=10+8(.5)+Σε ⇒ E(y₈)=14
Var(y₈)=8(2)=16
Δyₜ=.5+εₜ ⇒ E(Δy)=.5, Var(Δy)=2

RW variance=tσ²stationary limitVar
A random walk’s growing variance violates stationarity; first differencing removes the unit root.

M · 20-second method chooserfast recall

If asked…Write first
derive estimatordensity/model + dimensions
test restrictionsH₀, q, restricted/unrestricted
binary effectindex z, then p, then effect
assess classifierTP/FP/FN/TN with denominators
identify AR/MAACF vs PACF cut-off/decay
forecastbase, horizon, data vs forecast lags
nonstationaritymoment depending on t or unit root; then difference

Unit-root answer plan

  1. State weak-stationarity conditions.
  2. Back-substitute or solve characteristic equation.
  3. Show the mean/variance depends on t or a root equals one.

Information-criterion audit

https://www.asksia.ai/cheatsheet/manchester-econ31031 · Full ECON31031 study guide →
Forward this to your tutorial group · Shared by ________
AskSiav2 · 2026-10-19 · side 2/2
Two-hour on-campus invigilated examination in January 2027 (check the date in the university exam schedule).
Your exam allows one HANDWRITTEN A4 (both sides): copy side 1 by hand; do not bring this printout.

Want one for YOUR exact syllabus?

Sia is your free desktop study agent. Drop your The University of Manchester ECON31031 slides — Sia builds a sheet tailored to YOUR exam. Better than this library because it knows YOUR materials.

↓ Download Sia · Free