Regression modelEconometricsModel

White Test for Heteroskedasticity

Also known as: White's general heteroskedasticity test, White değişen varyans testi

OriginatorHalbert WhiteYear1980Sources1Related methods7

The White test, introduced by Halbert White in 1980, is a general test for heteroskedasticity that makes no assumption about its functional form. It regresses the squared OLS residuals on the regressors, their squares, and their cross-products, so it can detect heteroskedasticity related to any of these terms. The same 1980 paper introduced the heteroskedasticity-consistent ('White') standard errors that are the standard remedy when the test rejects.

Key highlights

  • Makes no assumption about the functional form of the heteroskedasticity, detecting a broad class of departures.
  • Requires no prior knowledge of which variables drive the variance.
  • Comes paired with the heteroskedasticity-consistent standard errors that remedy a rejection.
  • A standard, widely implemented general specification check.

Intuition

This section is available to Pro members. Upgrade to Pro

How it works

This section is available to Pro members. Upgrade to Pro

When to use it

Use the White test when you want a general check for heteroskedasticity without committing to a particular form, especially when you have no prior idea which variables drive the variance. Its breadth is also its cost: with many regressors the auxiliary regression consumes many degrees of freedom, reducing power, so for a small number of suspected variance drivers the more focused Breusch-Pagan test can be more powerful. Because the White test can also pick up general misspecification (its terms overlap with a functional-form check), a rejection should be interpreted alongside other diagnostics. When heteroskedasticity is confirmed, report White (heteroskedasticity-consistent) standard errors or estimate by weighted least squares.

Strengths & limitations

Strengths
  • Makes no assumption about the functional form of the heteroskedasticity, detecting a broad class of departures.
  • Requires no prior knowledge of which variables drive the variance.
  • Comes paired with the heteroskedasticity-consistent standard errors that remedy a rejection.
  • A standard, widely implemented general specification check.
Limitations
  • Consumes many degrees of freedom when there are many regressors, lowering power.
  • Less powerful than Breusch-Pagan against a specific, known form of heteroskedasticity.
  • A rejection can reflect general misspecification rather than heteroskedasticity alone.
  • Detects the problem but does not reveal the variance function or how to weight.

Common pitfalls

This section is available to Pro members. Upgrade to Pro

Applications

This section is available to Pro members. Upgrade to Pro

Frequently asked

How is the White test different from Breusch-Pagan?

Breusch-Pagan tests for heteroskedasticity tied to a chosen set of variables (often the regressors entered linearly), giving it focus and power against that form. The White test adds the squares and cross-products of the regressors, making it agnostic about the form but spreading its power thinner and consuming more degrees of freedom.

What do I do when the White test rejects?

Heteroskedasticity leaves OLS coefficients unbiased but their standard errors invalid. The standard remedy — introduced in the same White (1980) paper — is to report heteroskedasticity-consistent (robust) standard errors; weighted least squares is an alternative if you can model the variance.

Why might the White test have low power?

Because it includes many auxiliary terms (levels, squares, cross-products), its chi-square degrees of freedom grow quickly with the number of regressors. Spreading the test across many directions reduces its power against any single form, which is why a focused Breusch-Pagan test can do better when you know what drives the variance.

Sources

  1. 1.
    White, H. (1980). A heteroskedasticity-consistent covariance matrix estimator and a direct test for heteroskedasticity. Econometrica, 48(4), 817–838.

You have read it. What now?

Cite this page

ScholarGate. (2026, June 2). White Test. ScholarGate. https://scholargate.app/econometrics/white-test