Econometrics from First Principles
OLS, identification, and the toolkit that turns data into causal claims you can defend. Twelve modules from the linear regression to instrumental variables, panel data, and time series — with the assumptions, the failure modes, and the recipes for spotting bad inference in published work.
12
Modules
~11h 10m
Reading time
Intermediate
Level
Self-paced
Format
Syllabus
- 01→
What econometrics is for
Why we don't just compute correlations: the gap between description, prediction, and causation, and what each requires.
~40 minModule 01 - 02→
The linear regression — derivation and intuition
OLS as the line that minimises squared residuals. Geometric, algebraic, and projection-based views. When does the line mean something?
~60 minModule 02 - 03→
OLS assumptions and what they buy you
Linearity, exogeneity, homoskedasticity, no autocorrelation, normality. Which ones matter for unbiasedness, which for inference, which can you relax.
~60 minModule 03 - 04→
Standard errors, p-values, and confidence
What a p-value actually says (and doesn't), heteroskedasticity-robust SEs, clustered SEs, and the bootstrap.
~55 minModule 04 - 05→
Multiple regression and partial effects
The Frisch-Waugh-Lovell theorem, controlling vs collider bias, and why 'controlling for X' is rarely as clean as it sounds.
~60 minModule 05 - 06→
Endogeneity — the central problem
Omitted variables, simultaneity, measurement error, and reverse causation. The four ways your coefficient ends up wrong.
~55 minModule 06 - 07→
Instrumental variables
Exclusion restrictions, relevance, weak instruments, 2SLS in one slide. Why a good instrument is worth more than ten extra controls.
~65 minModule 07 - 08→
Difference-in-differences
Parallel trends, two-way fixed effects, the staggered treatment problem, and event-study plots that don't lie.
~55 minModule 08 - 09→
Panel data — fixed and random effects
When fixed effects beat random effects, when neither saves you, and the within-transformation that demystifies the algebra.
~55 minModule 09 - 10→
Time series — stationarity and AR/MA
Why you can't just run OLS on monthly data, the unit-root problem, and a primer on ARIMA, cointegration, and what 'long-run' means.
~65 minModule 10 - 11→
Limited dependent variables
Logit, probit, tobit, and the linear probability model — when each is right, and the marginal-effect interpretation that everyone gets wrong.
~50 minModule 11 - 12→
Reading empirical economics critically
A checklist for evaluating any published regression: identification strategy, robustness, falsification tests, and the questions referees should have asked.
~50 minModule 12
How to use this course
Start with module 01 if the material is new; skip ahead if you have prior exposure. Each module is self-contained but the arc is sequential — the projects in the final module assume the toolkit from modules 1-11. Every module ends with key takeaways and a curated further-reading list with primary sources.