The Basics
- Simple definition: A situation where the variability of a variable differs across the range of values of a predictor variable.
- Core idea: Inconsistent spread of data points in regression analysis.
- Think of it as: The scatter in your data gets wider or narrower as you move along the x-axis.
What It Actually Means
In econometrics, heteroscedasticity violates the assumption that errors have constant variance (homoscedasticity). It often appears in cross-sectional data – for example, predicting consumption across different income levels: low-income households have similar spending patterns (low variance), while high-income households vary widely (high variance). Heteroscedasticity doesn’t bias coefficients but makes standard errors wrong, leading to incorrect hypothesis tests.
Example
Studying household consumption across Pakistan, you’d find poor households spend similarly on basics, while rich households vary greatly – some save, some spend lavishly. This heteroscedasticity must be corrected for reliable results.
Why It Matters
Ignoring heteroscedasticity can lead to wrong conclusions about statistical significance. Economists use robust standard errors or transformations to address it.
See also
Econometrics • Regression Analysis • Homoscedasticity • Autocorrelation • Robust Standard Errors
Read more about this with MASEconomics:
How to Detect and Correct Heteroscedasticity in Econometric Models