Mostly Harmless Econometrics
Summary
A practical guide to the core methods of applied econometrics, focused on causal inference. The book covers regression, instrumental variables, differences-in-differences, regression discontinuity, quantile regression, and inference issues — emphasizing conceptual robustness over model-dependent assumptions.
Core Toolkit
The authors identify three essential tools for applied econometricians:
- Regression — designed to control for variables that may mask causal effects
- Instrumental variables — for analyzing real and natural experiments
- Differences-in-differences — using repeated observations to control for unobserved omitted factors
Four Research FAQs
Every empirical project should answer these questions:
- What is the causal relationship of interest?
- What experiment could ideally capture this causal effect?
- What is your identification strategy?
- What is your mode of statistical inference?
Structure
| Part | Chapters | Topics |
|---|---|---|
| I — Introduction | Ch 1-2 | Questions about Questions, The Experimental Ideal |
| II — The Core | Ch 3-5 | Regression, IV, DD & Panel Data |
| III — Extensions | Ch 6-8 | RD, Quantile Regression, Standard Errors |
Key Principles
- The CEF is the central object; regression approximates it
- Estimators in common use have simple, robust interpretations that are not heavily model-dependent
- If the estimates you get are not what you want, the fault lies in the econometrician, not the econometrics
- The book emphasizes finite-sample inference issues rather than asymptotic efficiency
See Also
- The Selection Problem — the fundamental challenge of causal inference
- Conditional Independence Assumption — the key assumption for causal regression
- Omitted Variables Bias — what goes wrong without proper controls
- BDA3 - Overview — Bayesian counterpart covering inference, regression, and model-based causal analysis
- Bayesian Workflow - Overview — iterative Bayesian approach to the same empirical questions