OLS can estimate an associative projection even when its structural-regression exogeneity condition fails. Identifying the intended structural slope requires valid moment conditions. In financial markets that condition can fail because price and quantity are jointly determined, omitted latent information affects both sides of a regression, or a measured regressor contains noise. This topic makes the failure explicit, then introduces instrumental variables and two-stage least squares as the route to a parameter identified from exogenous variation.
By the end of this note you should be able to:
- distinguish endogeneity from heteroscedasticity, serial correlation, and multicollinearity;
- explain why OLS is inconsistent when ;
- state relevance and exclusion as the two essential IV requirements;
- execute the conceptual two stages of 2SLS and interpret the fitted first-stage variation;
- identify weak instruments as a threat to inference rather than a minor diagnostic issue.
In this lesson
- Contents
- The big picture
The full lesson is for members.
Read the full lessonNew to SelfMadeQuant.com? See membership options or browse the Econometrics and Financial Time Series syllabus.