Lectures
EC614 — Time Series Econometrics & Macroeconomic Forecasting
One document per teaching week, in the order they are taught. Most are revealjs decks — press f for full screen, s for speaker notes, ? for all shortcuts. Every code chunk shown is the code that produced the output beside it.
Foundations
Week 1Course overview What macroeconomic forecasting is for, the data we will use, and how the course is assessed. Week 2Stationarity and non-stationarity Why stationarity is the hinge on which every later method turns; visual and statistical diagnosis. Week 3Non-stationary series Random walks, deterministic versus stochastic trends, differencing and over-differencing. Week 4Unit root tests: methods and problems ADF, PP, KPSS and ERS — what each actually tests, and why low power makes them easy to misread. Week 5Spurious regression Granger and Newbold’s warning, reproduced by simulation: high R², significant t, meaningless result.
Univariate modelling and forecasting
Week 6ARIMA modelling Box–Jenkins identification, estimation and diagnostic checking on bond returns and petrol prices. Week 7Forecast uncertainty and evaluation Interval forecasts, fan charts, and honest out-of-sample accuracy measures. Week 8Modelling volatility: ARCH and GARCH Volatility clustering in financial series, and forecasting the variance rather than the mean.
Multivariate systems and long-run relationships
Week 9Cointegration and VECM Engle–Granger and Johansen: when non-stationary series share a long-run equilibrium. Week 10ARDL and bounds testing Pesaran–Shin–Smith bounds testing with mixed I(0)/I(1) regressors, and the error correction form. Week 11Vector autoregressions Lag selection, Granger causality, impulse responses and forecast error variance decomposition. Week 12Structural VAR The identification problem, and what recursive and long-run restrictions really assume. Week 13SVAR and macroeconomic forecasting Putting structural VARs to work on policy scenarios and conditional forecasts.
Panel data
Looking for the hands-on version? Every method above has a matching walkthrough in the labs.