Econometrics
Econometrics applies statistics to economic and financial data. Learn its core tools, common problems like endogeneity and spurious results, and how traders use it.
Econometrics is the branch of economics that uses statistical methods to measure relationships in economic and financial data, test theories and make forecasts. Financial econometrics focuses on markets: asset returns, volatility, interest rates and risk. Many tools covered in this school, from regression and cointegration to GARCH and factor models, come from econometrics. Knowing its main problems and remedies helps traders interpret research and avoid drawing false conclusions from data.
Core tools#
| Tool | Purpose | Lesson |
|---|---|---|
| Linear regression | Estimate relationships between variables | Regression Analysis |
| Time series models (ARIMA) | Model persistence in a series | ARIMA |
| Volatility models (GARCH) | Model changing volatility | GARCH |
| Cointegration and error correction | Long run relationships | Cointegration |
| Vector autoregressions (VAR) | Joint dynamics of several series | Granger Causality |
| Panel data models | Data across many firms or countries over time | |
| Factor models | Explain returns with common factors | Factor Models |
| Event studies | Measure price reactions to events | Event-Driven Trading |
Event studies#
An event study measures abnormal returns around an event, such as an earnings announcement, merger or index addition:
abnormal return = actual return - expected return (from a model such as the market model)
cumulative abnormal return = sum of abnormal returns over an event windowCommon econometric problems#
| Problem | What it means | Remedy |
|---|---|---|
| Spurious regression | Unrelated trending series appear related | Use returns, test for cointegration. See Stationarity, Differencing and Unit Roots |
| Endogeneity | The explanatory variable is influenced by the outcome or by omitted factors | Instrumental variables, natural experiments |
| Omitted variable bias | A missing variable drives both sides | Include relevant controls |
| Heteroskedasticity | Error variance changes | Robust standard errors |
| Autocorrelation | Errors correlated over time | Newey West standard errors. See Autocorrelation and Partial Autocorrelation |
| Multicollinearity | Explanatory variables highly correlated | Combine or drop variables |
| Look ahead bias | Using unavailable data | Point in time data. See Look-Ahead Bias |
| Data snooping | Testing many models on the same data | Out of sample tests, corrections. See P-Hacking and Multiple Testing |
Correlation vs causation#
Econometrics tries to identify causal effects, not just correlations. In markets, this is especially hard: almost everything is connected, and traders react to the same information. Natural experiments, such as sudden rule changes or index inclusion decisions made for mechanical reasons, help isolate cause and effect.
Financial econometrics and Nobel prizes#
Several Nobel Memorial Prizes in Economic Sciences have honoured financial econometrics, including Robert Engle (volatility models) and Clive Granger (cointegration) in 2003, and Eugene Fama, Lars Peter Hansen and Robert Shiller in 2013 for empirical analysis of asset prices. Hansen developed the generalised method of moments (GMM), widely used to test asset pricing models.
Using econometrics as a trader#
- Ask what the model assumes and whether the data fits those assumptions.
- Check standard errors are adjusted for autocorrelation and heteroskedasticity.
- Be sceptical of high R² with trending data.
- Prefer out of sample evidence to in sample fit.
- Read research critically: sample period, data sources and number of tests matter. See Reading Academic Papers.
Frequently asked questions#
What is econometrics?#
The application of statistical methods to economic and financial data to measure relationships, test theories and make forecasts.
What is an event study?#
A method that measures abnormal returns around an event, such as earnings or mergers, by comparing actual returns with those expected from a model.
What is endogeneity?#
A problem where the explanatory variable is correlated with the error term, often because of omitted factors or two way causation, biasing estimates.
You have finished the Maths and Statistics track. Continue with research methods, starting with The Trading Research Process.
3 quick questions on this lesson. Get them all right to finish it.
Turn on JavaScript to take the quiz.
Mentioned in
- ARIMAMath and Statistics
- OptimizationMath and Statistics