ECON31031 Advanced Econometrics
ECON31031 Overview
- The University of Manchester
- Semester 1 2026/27
- 9 chapters
- Economics
ECON31031 Advanced Econometrics is a Level 3 module worth 20 credits that extends the second-year module ECON20110 Econometrics, its compulsory prerequisite, in two directions.
- Assessed by A two-hour January examination worth 75% and an online coursework in R worth 25%.
- Exam shape Three answers: one of two microdata questions and two of three time series questions.
- Core method Maximum likelihood links the binary models in Part I with ARMA estimation in Part II.
- How to prepare Practise derivations and calculator work by hand, then write a one-line interpretation.
What you get in the full ECON31031 Exam Bible
- 9 chapters, 43 pages — every examinable topic
- 26 original practice questions, most with model answers in the Final Exam format
- 9 worked examples, solved the way ECON31031 marks
- 14 computed figures, redrawn from the course’s own numbers
- Exam-morning plan for the Final Exam (75%) — what to revise last, where marks are lost
Free on this page: assessment table, chapter map, one worked example, key terms, FAQ. The rest is in the full exam bible.
How ECON31031 is assessed
| Component | Weight | Format |
|---|---|---|
| Online Coursework | 25% | A single coursework covering Part I of the syllabus: an empirical microdata application in R. |
| Final Exam | 75% | On-campus invigilated exam of two hours in January. Answer three questions: one of two in Section A on microdata and two of three in Section B on macrodata. One handwritten A4 notes sheet and a scientific calculator that cannot store text are allowed. |
What other students are asking about ECON31031
- Where can I find lecture notes or study guides for ECON31031 at manchester?
- The ECON31031 study guide in AskSia Library covers Advanced Econometrics in 9 chapters, with the assessment table, a chapter-by-chapter topic map, key terms and a worked example, free to read on this page.
- Is there a final exam in ECON31031, and how much is it worth?
- Yes. In the offering this AskSia Library guide is built from, Final Exam carries 75% of the ECON31031 mark; the full split is Online Coursework 25%, Final Exam 75%.
- What does ECON31031 cover on maximum likelihood estimation for iid data?
- In AskSia Library's ECON31031 guide, the Maximum Likelihood Estimation for IID Data chapter explains likelihood and log-likelihood, the score, closed-form and numerical solutions, Hessian-based covariance estimators and likelihood ratio tests.
- What does ECON31031 cover on binary response models: lpm, logit and probit?
- In AskSia Library's ECON31031 guide, the Binary Response Models: LPM, Logit and Probit chapter explains bernoulli responses, the linear probability model and its problems, logit odds, the generalised linear model view, latent variables and limited dependent variables.
- What does ECON31031 cover on marginal effects and odds ratios?
- In AskSia Library's ECON31031 guide, the Marginal Effects and Odds Ratios chapter explains derivatives for continuous regressors, discrete changes for dummies, effects at the average and average effects, delta-method standard errors and odds ratios.
Current ECON31031 dates
| Date | Item | Control |
|---|---|---|
| January 2027 | Final examination period | The exam page refers to the January paper; the exact date appears in the university examination schedule. |
Dates are as published in the Semester 1, 2026/27 exam information page. Confirm exact deadlines and submission settings in the live LMS.
What ECON31031 covers
Nine chapters follow the two parts of the syllabus: least squares and maximum likelihood, binary response models and their assessment, then stationary time series, ARMA forecasting and unit roots.
OLS and the Linear Model in Matrix Form
The standard linear model in matrix notation, Gauss-Markov assumptions, projection and residual-maker matrices, the variance estimator, t, F and auxiliary-regression tests.OLS and the Linear Model in Matrix Form restates the standard linear model as y = Xβ + ε so that every later estimator has a benchmark. With n observations and k regressors plus an intercept, X has n rows and k+1 columns.
- The standard linear model y = Xβ + ε and its dimensions
- Gauss-Markov assumptions and the BLUE property
- The OLS estimator, its mean and its conditional variance
- Projection and residual-maker matrices
- Trace rules and the unbiased estimator s²
- Testing with t, F and restricted least squares
Maximum Likelihood Estimation for IID Data
Likelihood and log-likelihood, the score, closed-form and numerical solutions, Hessian-based covariance estimators and likelihood ratio tests.Maximum Likelihood Estimation for IID Data is the method behind most of the module. For an independent sample, the likelihood is the product of individual densities, and its logarithm is a sum that can be differentiated.
- Likelihood and log-likelihood for IID samples
- The score and first-order conditions
- Closed-form and numerical solutions
- The Hessian and covariance estimation
- Likelihood ratio tests against chi-squared values
- Normal linear models: ML versus OLS variance estimators
Binary Response Models: LPM, Logit and Probit
Bernoulli responses, the linear probability model and its problems, logit odds, the generalised linear model view, latent variables and limited dependent variables.Binary Response Models treat a yes or no outcome as a Bernoulli variable whose success probability depends on individual characteristics through a distribution function of the linear index.
- Binary, ordinal and multinomial responses
- Bernoulli probabilities that vary across individuals
- The linear probability model and its two problems
- White's heteroscedasticity-consistent covariance matrix
- Logit, log odds and odds ratios
- Generalised linear models and link functions
Marginal Effects and Odds Ratios
Derivatives for continuous regressors, discrete changes for dummies, effects at the average and average effects, delta-method standard errors and odds ratios.Marginal Effects and Odds Ratios answer the question every applied binary model raises: how much does a characteristic change the probability? In the linear probability model the coefficient is the effect.
- Why coefficients are not effects in nonlinear models
- Marginal effects for continuous regressors
- Discrete changes for dummy variables
- Effects at the average versus average marginal effects
- Delta-method standard errors
- Odds ratios and percentage-point interpretation
Binary Model Accuracy, ROC Curves and Link Tests
Hit and miss tables, Kuipers, precision, recall, F1, pseudo-R-squared measures, the Brier score, ROC curves, AUC, the phi coefficient and the link test.Binary Model Accuracy, ROC Curves and Link Tests covers how to judge a fitted binary model.
- In-sample fitted values and out-of-sample prediction
- Hit and miss tables at a cut-off
- Hit count, Kuipers score, precision, recall and F1
- Pseudo-R-squared measures and the Brier score
- ROC curves and the AUC by the trapezium rule
- Imbalanced outcomes and the phi coefficient
Time Series Foundations and Stationarity
Univariate time series, lag and difference operators, autocovariance, sample autocorrelation, correlograms, white noise and weak stationarity.Time Series Foundations and Stationarity opens Part II of the module, which follows on from the time series material in the prerequisite econometrics module.
- Univariate time series and sampling frequency
- Lag operators and lag polynomials
- First, seasonal and second differences
- Autocovariance, autocorrelation and the sample ACF
- Time plots, correlograms and the PACF
- White noise
AR, MA and ARMA Processes
AR(1) and AR(p) moments, MA(q) cut-offs and invertibility, the PACF, identification rules, Wold's theorem, ARMA models and AR(2) stationarity conditions.AR, MA and ARMA Processes are the stationary models of Part II.
- AR(1) definition and stationarity condition
- Mean, variance and ACF of an AR(1)
- AR(p) processes and lag polynomials
- MA(q) moments, cut-offs and invertibility
- The PACF and the eyeball identification rules
- Wold's theorem and ARMA(p,q)
ARMA Forecasting, Estimation and Model Selection
Forecast horizons and bases, conditional and unconditional forecasts, AR and MA forecast recursions, maximum likelihood for ARMA, AIC and BIC.ARMA Forecasting, Estimation and Model Selection turns fitted models into forecasts and chooses between them. The vocabulary matters: in-sample and out-of-sample data, extrapolation, the forecast base and the horizon.
- Forecast vocabulary: in-sample, out-of-sample, base and horizon
- Conditional and unconditional expectations
- Multi-step AR(1) and AR(p) forecasts
- MA(q) forecasts and mean reversion
- Maximum likelihood estimation of ARMA models
- AIC and BIC model selection
Unit Roots and Nonstationary Processes
Random walks, unit roots in AR(1) and AR(2) processes, variance growth, the differenced form, integrated series, ARIMA and the step to VARs.Unit Roots and Nonstationary Processes cover what happens when stationarity fails in the most consequential way.
- The AR(1) with coefficient one: the random walk
- Why variance grows and the mean does not settle
- Unit roots in higher-order autoregressions
- The differenced form of an autoregression
- Integrated series and ARIMA models
- The step to vector autoregressions
Part I develops models for microdata, particularly responses limited to a few values, and the method that underpins them: parametric maximum likelihood for independent and identically distributed data.
Starting from the standard linear model written in matrix form, it builds log-likelihoods, scores and likelihood ratio tests, then applies them to the linear probability model, logit and probit, marginal effects, odds ratios and a family of accuracy measures that ends with ROC curves and the link test for misspecification.
Part II turns to macroeconomic time series.
It defines white noise and weak stationarity, introduces the lag and difference operators, and works through autoregressive, moving average and ARMA processes, their autocorrelation patterns, forecasting and model selection by information criteria. The syllabus then moves to unit roots and closes with an introduction to vector autoregressions.
Both parts use R, with worksheets on linear models, generalised linear models and time series data, and the core text is Verbeek’s A Guide to Modern Econometrics.
The assessment has two components. A single online coursework worth 25% asks for an empirical microdata application in R on Part I of the syllabus. The final examination is worth 75% and takes place on campus in the January examination period.
The module’s exam page describes a two-hour invigilated paper in which you answer three questions: one of two microdata questions in Section A and two of three macrodata questions in Section B. One handwritten A4 sheet of notes is allowed, both sides usable, and a scientific calculator that cannot store text is needed for some questions.
An example paper based on last year’s exam is due near the end of teaching, with the format changed for this year.
This guide follows that structure.
Each chapter explains the method, draws a figure from its own numbers, works a complete example and finishes with short practice questions, so that derivations and calculations can be rebuilt by hand under exam conditions.
ECON31031 cheatsheet for the final exam
AskSia publishes a free ECON31031 cheatsheet for the final exam, a handwritten A4 planner that follows the official permitted-materials rule for this assessment and condenses the topics in this guide.
Worked example · free
Turn a fitted logit into probabilities, odds and an odds ratio
- 2Build the two indices. With the course: -0.8 + 0.06(15) + 0.7 = 0.8. Without it: -0.8 + 0.9 = 0.1.
- 2Apply the logistic function. With the course e^0.8/(1 + e^0.8) = 2.2255/3.2255 = 0.690; without it e^0.1/(1 + e^0.1) = 1.1052/2.1052 = 0.525. The course raises the probability by about 0.165 for this student.
- 2The odds are 2.2255 and 1.1052, so the odds ratio is 2.014, which equals e^0.7. The odds ratio is the same for every student, while the change in probability depends on the starting index.
Key terms
- Maximum likelihood estimator
- The parameter value that maximises the probability the model assigns to the observed sample.
- Score function
- The vector of first derivatives of the log-likelihood with respect to the parameters.
- Likelihood ratio test
- A test that compares twice the gap between unrestricted and restricted log-likelihoods with a chi-squared distribution.
- Linear probability model
- A binary response model in which the probability of success is a linear function of the regressors, estimated by OLS.
- Logit model
- A binary response model that uses the logistic distribution function, so the log odds are linear in the regressors.
- Probit model
- A binary response model that uses the standard normal distribution function to map the linear index into a probability.
- Marginal effect
- The change in the fitted probability of success caused by a small change in one regressor, holding the others fixed.
- Odds ratio
- The factor by which the odds of success are multiplied when a regressor rises by one unit in a logit model.
- ROC curve
- A plot of the true positive rate against the false positive rate across a range of classification cut-offs.
- White noise
- A process with zero mean, constant variance and no correlation between values at different dates.
- Weak stationarity
- The property of a process whose mean, variance and autocovariances are finite and do not change with the date.
- Partial autocorrelation function
- The sequence of last coefficients from autoregressions of increasing order, used to identify the order of an AR process.
- Unit root
- A root of the autoregressive characteristic equation equal to one, which makes a process nonstationary, as in a random walk.
ECON31031 FAQ
How hard is Advanced Econometrics?
The challenge is concentration and algebra rather than volume. The final examination carries 75% of the result in a two-hour paper of three questions, and the module assumes the second-year econometrics prerequisite.
Questions ask for derivations such as log-likelihoods and scores as well as calculations with a scientific calculator, and the outline stresses the pitfalls of applied modelling, so careful set-up matters as much as arithmetic.
How is the module assessed?
There are two components. An online coursework worth 25% asks for an empirical application to microdata using R, set on Part I of the syllabus. A final examination worth 75% follows in the January examination period. The course outline lists these weights, and the exam information page adds the format details for the paper itself.
What is the format of the January exam?
The exam page describes an on-campus invigilated paper lasting two hours. You answer three questions: one from Section A, chosen from two questions on the microdata material, and two from Section B, chosen from three on the macrodata material. The course outline was written before the format was settled, so treat the exam page and the printed rubric as the guide.
Can I bring notes or a calculator into the exam?
Yes, within limits. You may bring one handwritten A4 sheet of notes, and both sides can be used. A scientific calculator is needed to complete some questions, but calculators that can store text are not permitted, in line with the economics calculator policy. Check the rubric on the day, because it states the rules that apply.
Are past exam papers available for this module?
The exam page says an example paper based on last year’s examination will be released near the end of the teaching period, and that the format has changed for this year. It was not yet posted in the materials this guide draws on, so the practice questions here are original and modelled on the exercise sheets.
Which software does the module use?
Both parts use R, usually through RStudio. The worksheets move from exploratory analysis and linear models with lm() to generalised linear models with glm() for binary data, and then to time series analysis. Video walkthroughs accompany the worksheets, and the module suggests attempting each worksheet before watching the walkthrough.
What should I revise first before the exam?
Secure the shared tools first: matrix OLS, log-likelihoods and the likelihood ratio test, because both sections lean on them. Then give time series extra weight, since two of your three answers come from Section B. Finish with binary-model calculations such as probabilities, marginal effects and accuracy measures, which make reliable Section A material.
Questions students ask in each ECON31031 chapter
- Why does s² divide by n - k - 1?
- The expected residual sum of squares equals the trace of the residual maker times the error variance, and that trace is n - k - 1. Dividing by this number makes the estimator unbiased, while dividing by n would understate the error variance. Chapter 01 · OLS and the Linear Model in Matrix Form
- When does a maximum likelihood estimator have a closed form?
- It has one when the score equations can be solved algebraically, as for a Bernoulli proportion or the slopes of a normal linear model. When parameters enter through a nonlinear function summed over observations, such as a probit, the equations must be solved numerically. Chapter 02 · Maximum Likelihood Estimation for IID Data
- Why not just use OLS for a binary outcome?
- OLS on a binary outcome is the linear probability model. It is easy to estimate, but it can produce fitted probabilities above one or below zero, forces each characteristic to have a constant effect, and has heteroscedastic errors by construction, so its usual standard errors are biased. Chapter 03 · Binary Response Models: LPM, Logit and Probit
- Where is a logit marginal effect largest?
- It is largest where the fitted probability is one half, because the logistic density equals p(1 - p) and peaks there at 0.25. The same coefficient therefore moves probabilities less for individuals whose probabilities are already close to zero or one. Chapter 04 · Marginal Effects and Odds Ratios
- Why can a high hit count be misleading?
- When one outcome dominates, a model that predicts the common outcome for almost everyone scores a high hit count while separating the groups badly. Kuipers, precision, recall and the AUC reveal that weakness, as the benefits example in the notes shows. Chapter 05 · Binary Model Accuracy, ROC Curves and Link Tests
- What makes a process weakly stationary?
- Three conditions must hold at every date: the mean is finite and constant, the variance is finite and constant, and the covariance between two values depends only on how many periods separate them. White noise is stationary with every autocorrelation beyond lag zero equal to zero. Chapter 06 · Time Series Foundations and Stationarity
- How do I tell an AR from an MA in a correlogram?
- Use the eyeball rules from the notes. An AR(p) has an ACF that decays and a PACF that cuts off after p lags. An MA(q) has an ACF that cuts off after q lags and a PACF that decays. Mixed patterns suggest an ARMA. Chapter 07 · AR, MA and ARMA Processes
- Why do AR forecasts converge to the mean?
- Each step multiplies the gap between the last forecast and the mean by the AR coefficient, which is smaller than one in absolute value for a stationary process. The gap therefore shrinks geometrically, so long-horizon forecasts approach the unconditional mean. Chapter 08 · ARMA Forecasting, Estimation and Model Selection
- What is a unit root?
- It is a root of the characteristic equation of an autoregression equal to one. In an AR(1) it means the coefficient is one, the random walk case, where the variance grows over time and the process is not stationary. Chapter 09 · Unit Roots and Nonstationary Processes
How to study for the exam
Organise revision around the three things the paper asks you to do: derive, calculate and interpret. For derivations, practise writing a density, the log-likelihood and the score from a blank page, and learn to say clearly when a closed-form estimator exists and why it does not when parameters sit inside a nonlinear function.
Rehearse the matrix results from the first weeks, the variance of the OLS estimator and the residual-maker identities, because they reappear in testing questions. For calculations, build fluency with probabilities from logit and probit indices, marginal effects, odds ratios, hit and miss tables, the AUC by the trapezium rule, AR and MA moments, multi-step forecasts and information criteria.
Use only a scientific calculator and the normal and chi-squared tables, as in the exam. For interpretation, finish every answer with a sentence that states the decision or the meaning. Because Section B supplies two of your three answers, schedule more time for stationarity, ARMA identification, forecasting and unit roots, but keep one Section A topic polished.
Build your handwritten notes sheet gradually: add a formula only after you have used it in a worked problem, and group it by section so you can find it under time pressure.
Your AI Economics tutor for ECON31031
Stuck on a hard ECON31031 question? Sia is AskSia’s AI Economics tutor — ask any ECON31031 Advanced Econometrics question and get a clear, step-by-step explanation grounded in how the course is actually taught and assessed. Read this whole study guide free, then take your hardest questions to Sia.