Learn Stata

NetCourse^®461: Univariate Time Series with Stata

Course length: 6 weeks (4 lessons)

Content:

Learn about univariate time-series analysis with an emphasis on the practical aspects most needed by practitioners and applied researchers. Written for a broad array of users, including economists, forecasters, financial analysts, managers, and anyone who wants to analyze time-series data. Become expert in handling date and date–time data, time-series operators, time-series graphics, basic forecasting methods, ARIMA, ARMAX, and seasonal models.

We provide lesson material, detailed answers to the questions posted at the end of each lesson, and access to a discussion board on which you can post questions for other students and the course leader to answer.

Prerequisites:

• Stata 18 installed and working
• Course content of NetCourse 101 or equivalent knowledge
• Familiarity with basic cross-sectional summary statistics and linear regression
• Internet web browser, installed and working

Note: This course is platform independent. You can use Stata on Windows, Mac or Linux.

Trainer: Miguel Dorta

Trainer: Gustavo Sánchez

Course content

Lesson 1: Introduction

• Course outline
• Follow along
• What is so special about time-series analysis?
• Time-series data in Stata
  • The basics
  • Clocktime data
• Time-series operators
  • The lag operator
  • The difference operator
  • The seasonal difference operator
  • Combining time-series operators
  • Working with time-series operators
  • Parentheses in time-series expressions
  • Percentage changes
• Drawing graphs
• Basic smoothing and forecasting techniques
  • Four components of a time series
  • Moving averages
  • Exponential smoothing
  • Holt–Winters forecasting

Lesson 2: Descriptive analysis of time series

• The nature of time series
  • Stationarity
• Autoregressive and moving-average processes
  • Moving-average processes
  • Autoregressive processes
  • Stationarity of AR processes
  • Invertibility of MA processes
  • Mixed autoregressive moving-average processes
• The sample autocorrelation and partial autocorrelation functions
  • A detour
  • The sample autocorrelation function
  • The sample partial autocorrelation function
• A brief introduction to spectral analysis—The periodogram

Lesson 3: Forecasting II: ARIMA and ARMAX models

• Basic ideas
  • Forecasting
  • Two goodness-of-fit criteria
  • More on choosing the number of AR and MA terms
• Seasonal ARIMA models
  • Additive seasonality
  • Multiplicative seasonality
• ARMAX models
• Intervention analysis and outliers
• Final remarks on ARIMA models

Note: There is a one-week break between the posting of Lessons 3 and 4; however, course leaders are available for discussion.

Lesson 4: Regression analysis of time-series data

• Basic regression analysis
• Autocorrelation
  • The Durbin–Watson test
  • Other tests for autocorrelation
• Estimation with autocorrelated errors
  • The Newey–West covariance matrix estimator
  • ARMAX estimation
  • Cochrane–Orcutt and Prais–Winsten methods
• Lagged dependent variables as regressors
• Dummy variables and additive seasonal effects
• Nonstationary series and OLS regression
  • Unit-root processes
• ARCH
  • A simple ARCH model
  • Testing for ARCH
  • GARCH models
  • Extensions

Note: The previous four lessons constitute the core material of the course. The following lesson is optional and introduces Stata’s multivariate time-series capabilities.

Bonus lesson: Overview of multivariate time-series analysis using Stata

• VARs
  • The VAR(p) model
  • Lag-order selection
  • Diagnostics
  • Granger causality
  • Forecasting
  • Impulse–response functions
  • Orthogonalized IRFs
  • VARX models
• VECMs
  • A basic VECM
  • Fitting a VECM in Stata
  • Impulse–response analysis