ECON 6376: Applied Time Series Econometrics
A graduate course in the George Washington University Department of Economics, offered in the Applied Economics MA program. Materials are posted here as the semester proceeds.
Syllabus
Syllabus (PDF) — including the tentative topic schedule and assignment due dates.
Modules
| # | Topic | Notes | Slides | Materials | Literature of Note |
|---|---|---|---|---|---|
| 1 | Intro: AR(1), ACF, Random Walk, Spurious Regression | Notes | Slides · PDF | AI pack | Papers (zip) |
| 2 | Testing for Stationarity | Notes | Slides · PDF | AI pack | Papers (zip) |
| 3 | AR(p), MA(q) and Their ACF/PACF Fingerprints | Notes | Slides · PDF | AI pack | — |
| 4 | ARMA Modeling, Estimation, and Information Criteria | Notes | Slides · PDF | AI pack | — |
| 5 | Diagnostics, Seasonality, and SARIMA | Notes | Slides · PDF | AI pack | Papers (zip) |
| 6 | Forecasting Fundamentals | Notes | Slides · PDF | AI pack | — |
| 7 | Forecast Evaluation and Combinations | Notes | Slides · PDF | AI pack | Papers (zip) |
| 8 | ARDL Models and Dynamic Multipliers | Notes | Slides · PDF | AI pack | — |
| 9 | Granger Causality | Notes | Slides · PDF | AI pack | Papers (zip) |
| 10 | Cointegration and Error Correction | Notes | Slides | AI pack | Papers (zip) · Murray (1994) |
| 11 | Structural VAR Foundations | Notes | Slides · PDF | AI pack | Papers (zip) |
| 12 | Impulse Response Functions | — | — | — | — |
| 13 | FEVD and VECMs | — | — | — | — |
The Literature of Note column collects optional readings — the original papers behind each module’s methods — zipped per module. None of them is required.
Problem Sets
- Problem Set 1 — Modules 1–2
- Problem Set 2 — Modules 3–5
- Problem Set 3 — Modules 5–7 (data: ps3_series.csv)
- Problem Set 4 — Modules 9–10 (data: ps4_series.csv)
Further problem sets are posted as they are assigned. Problem sets after the midterm work each method twice: once on transparent or simulated data, where the right answer is not in dispute, and once on the series you assembled for your own midterm.
Exams
- Midterm (take-home) — an applied project on four or more time series of your choosing. Posted now, due in October, so you can see from the first week what the first half of the course is building toward. The dataset and forecasts you produce are reloaded in Problem Sets 4 and 5.
The final is an in-person, closed-book exam held during the university final exam period. Everything you need — plots, regression output, test results — is printed in the exam itself. The Final exam blueprint (PDF) covers the topics, the question formats, and several fully worked sample questions.
Data and Software
- UNRATE.csv — cached FRED unemployment-rate series, so the notes reproduce without a FRED API key.
- UNRATENSA.csv — the same unemployment rate before seasonal adjustment, used throughout Module 5.
- GDPC1.csv — cached FRED quarterly real-GDP series, used by the Module 2 pack’s trend-vs-difference exploration.
- A939RX0Q048SBEA.csv — cached FRED quarterly real GDP per capita (BEA, chained 2017 dollars), the worked example that runs through the Module 6 slides and pack.
- ps3_series.csv — the custom monthly series (January 1990 to December 2019) used throughout Problem Set 3.
- ps4_series.csv — 500 observations of X, Y, and Z for Problem Set 4.
- ICSA.csv — cached weekly initial unemployment-insurance claims, seasonally adjusted, used in Modules 8 and 9.
- Consumption and income CSV · Data notes — historical quarterly consumption/income archive (1959Q1–2009Q3), used in Module 10.
- helpers.zip — the course R function library (simulators, testing, diagnostics, forecasting, VAR utilities). Unzip into your project folder and source what you need, e.g.
source(here::here("helpers", "simulators.R")). See the includedREADME.mdfor the full function inventory and which lecture each function is built in.
The course runs in R. You will need the fredr, forecast, urca, vars, ggplot2, quadprog, and here packages (quadprog is used by the Module 7 forecast-combination helpers). A FRED API key is needed only for live data pulls, not to reproduce the notes.
AI Learning Packs
Each pack is a self-contained folder for working through a module alongside an AI assistant. Download it, unzip it, and open the folder with whatever AI tool you use — it is set up so the assistant reads the module’s own notes and notation rather than improvising its own.
Inside you get a runnable lab, the module notes for grounding, the R functions the module builds, the cached data, and a set of executable checks. Start with START_HERE.md.
The packs contain no solutions, rubrics, or answer keys. The checks verify that the code and data behave the way the module says they do — so when you and an assistant disagree about a fact, there is a referee that neither of you controls. They cannot tell you whether your reading of a correlogram is any good, and they do not try.
AI assistance is permitted, with disclosure, on take-home graded work: the problem sets and the midterm. The final is an in-person, closed-book exam with no devices, so the packs are for preparing for it, not for use during it. What gets graded is understanding you can defend.
Interactive Tools
Browser-based training exercises for the course are on the Time Series Tools page.