Description

Chapter 6

These slides accompany Chapter 6 (“Heteroskedasticity and Autocorrelation Robust (HAR) Inference”) of the book

Yongmiao Hong, Oliver Linton, Jiajing Sun
Econometrics and Time Series Methods: Theory, Applications, and R Implementation.

The slides provide a detailed treatment of inference in regression and time series models when the error terms may exhibit heteroskedasticity, serial correlation, or both. They contrast conventional (homoskedastic, serially uncorrelated) inference with robust approaches, and motivate why HAR methods are essential in empirical econometrics. Core topics typically include heteroskedasticity-robust covariance estimators, heteroskedasticity and autocorrelation consistent (HAC) estimators, long-run variance estimation, bandwidth and kernel choices, and the construction of robust t- and Wald-type tests. The slides also discuss practical issues such as finite-sample performance, the consequences of misspecifying the error structure, and how robust procedures modify standard regression output and interpretation.

In line with the book’s integration of theory, applications, and computation, the slides show how to implement HAR inference in R, including the computation of robust standard errors, test statistics, and confidence intervals in both cross-sectional and time series settings. They are suitable for advanced undergraduate and graduate teaching, and can be directly used or adapted by instructors for lectures, tutorials, or self-study.

Unless otherwise indicated, the slides are shared under the Creative Commons Attribution–NonCommercial 4.0 International License (CC BY-NC 4.0). Readers and instructors who wish to request the LaTeX source files or provide feedback are welcome to contact us at jiajing.sun@gmail.com.