Structural Break Hausman Test
Hausman Specification Test with Structural Break Correction · Also known as: Hausman test under structural change, structural change Hausman specification test, break-robust Hausman test, panel specification test with breaks
The Structural Break Hausman Test extends the classical Hausman (1978) specification test to panel or time-series settings where the data-generating process shifts at one or more break points. By detecting structural breaks first and then running the Hausman comparison within each regime, researchers can reliably choose between fixed effects and random effects estimators even when the underlying relationship changes over time.
Read the full method
Sign in with a free account to read this section.
Method map
The neighbourhood of related methods — select a node to explore.
When to use it
Use the structural break Hausman test when you have panel data spanning a period likely to contain policy changes, economic crises, or other regime shifts, and you need to decide between fixed and random effects within each stable sub-period. It is especially appropriate when a preliminary break test rejects parameter constancy. Do not use the standard Hausman test without break correction if the Bai-Perron or Zivot-Andrews tests signal a break, as the pooled Hausman statistic will be distorted. Not suitable for very short panels (T < 10 per regime) because the within-regime sample may be too small for reliable asymptotic inference.
Strengths & limitations
- Prevents misspecification of the FE vs. RE choice caused by ignoring regime changes in the panel.
- Allows regime-specific model selection, which is more credible than a single pooled specification test.
- Builds directly on the well-understood Hausman (1978) framework, so the test statistic and its chi-squared distribution are familiar to practitioners.
- Compatible with standard panel software when break dates are imposed as known after a prior break test.
- Improves efficiency of coefficient estimates by using RE where it is valid within a stable regime.
- Requires pre-testing for break dates, which introduces pre-test bias: if the break is mislocated, the within-regime Hausman test may still be misleading.
- Each sub-sample has fewer observations, reducing the power of the Hausman test — particularly problematic with few time periods per unit.
- Multiple sequential tests (one per regime) inflate the overall type I error rate unless appropriate corrections are applied.
- Does not detect cross-sectional heterogeneity in break timing (each unit may break at a different point).
- The test is asymptotic; in small panels the chi-squared approximation may be poor.
Frequently asked
Do I need to know the break date in advance?
No. You first run a break-detection test (e.g., Bai-Perron or Zivot-Andrews) to estimate the break date, then split the sample at that date and apply the Hausman test within each sub-sample. The sequentiality is the standard practice, even though it introduces pre-test dependence.
What if the Hausman test gives opposite conclusions in the two regimes?
That is a meaningful finding: it suggests the random-effects assumption holds in one period but not the other. The appropriate response is to use regime-specific estimators — RE for the stable regime and FE for the regime where RE is rejected — rather than a single pooled model.
How is this different from just including a structural break dummy in the standard Hausman test?
A break dummy shifts the intercept but does not change the slope coefficients or the within-regime covariance structure used by the Hausman statistic. Splitting the sample at the break date allows all coefficients and error variances to differ across regimes, which is the correct treatment when a full structural change has occurred.
Can I apply this with more than one break?
Yes. The Bai-Perron procedure identifies multiple breaks, producing three or more sub-samples. A separate Hausman test is run within each sub-sample. The power of each test decreases as more breaks are added because sub-samples become smaller.
Does this test require a balanced panel?
No, but unbalanced panels complicate the within-regime sample sizes. With an unbalanced panel, ensure that each regime has sufficient observations per unit to sustain the FE and RE estimators before running the Hausman comparison.
Sources
- Hausman, J. A. (1978). Specification tests in econometrics. Econometrica, 46(6), 1251–1271. DOI: 10.2307/1913827 ↗
- Perron, P. (2006). Dealing with structural breaks. In T. C. Mills & K. Patterson (Eds.), Palgrave Handbook of Econometrics, Vol. 1 (pp. 278–352). Palgrave Macmillan. link ↗
How to cite this page
ScholarGate. (2026, June 3). Hausman Specification Test with Structural Break Correction. ScholarGate. https://scholargate.app/en/econometrics/structural-break-hausman-test
Which method?
Set this method beside its closest kin and read them side by side — the library lays the books on the table; the choice is yours.
- Fixed Effects ModelEconometrics↔ compare
- Panel Hausman TestEconometrics↔ compare
- Structural Break Fixed Effects ModelEconometrics↔ compare
- Structural Break Random Effects ModelEconometrics↔ compare
- Zivot-Andrews Structural Break TestEconometrics↔ compare