# Heteroskedasticity

*Math & Statistics — Finicade finance glossary*

Heteroskedasticity means the variance of regression errors is not constant across observations — it wrecks standard errors, not the coefficients.

Heteroskedasticity means the variance of the errors is not constant across observations — spending varies more among high earners, volatility varies across market regimes. It doesn't bias OLS coefficients, but it wrecks the standard errors, so significance tests become unreliable. The standard fix is not to abandon OLS but to use heteroskedasticity-robust standard errors, which is now the default in most applied work.

**Also known as:** heteroscedasticity, non-constant variance, robust standard errors

**Related terms:** [Residual](https://finicade.com/glossary/residual), [Ordinary Least Squares](https://finicade.com/glossary/ordinary-least-squares), [Standard Error](https://finicade.com/glossary/standard-error), [Volatility Clustering](https://finicade.com/glossary/volatility-clustering), [GARCH](https://finicade.com/glossary/garch)

**Taught in:** Regression Range — Inference Exam

Source: https://finicade.com/glossary/heteroskedasticity