An OLS slope in a bid--ask-spread regression can look precise while being economically meaningless if trading volume and liquidity are jointly determined. The failure is not a standard-error problem: when a regressor is correlated with the error, OLS converges to the wrong coefficient even in an arbitrarily large sample. This note identifies the main routes to endogeneity and separates them from the other regression failures diagnosed earlier in the unit.
Learning objectives. By the end of this note you should be able to:
- define endogeneity as regressor--disturbance correlation;
- show the omitted term in the probability limit of OLS;
- distinguish simultaneity, omitted variables, and measurement error as sources;
- explain why robust standard errors cannot repair endogeneity;
- formulate the finance intuition for a simultaneous price, spread, or volume system.
In this lesson
- The load-bearing OLS condition
- Three routes to correlation
- What endogeneity is not
Before you start
The full lesson, with worked examples and exercise solutions, is for members.
Read the full lessonNew to SelfMadeQuant.com? See membership options or browse the Econometrics and Financial Time Series syllabus.