ScholarGate
Get started
Regression modelEconometricsEconometrics / time seriesModel

Panel Zivot-Andrews Structural Break Unit Root Test

Also known as: panel ZA test, panel structural break unit root test, Zivot-Andrews panel unit root test, panel endogenous break unit root test

OriginatorZivot & Andrews (1992); extended to panel settings by subsequent literatureYear1992 (panel extension: 2000s)Sources2Related methods7

The Panel Zivot-Andrews test extends the single-series Zivot-Andrews (1992) structural break unit root test to panel data, allowing each cross-sectional unit to have its own endogenously determined break date. It tests the null of a unit root against the alternative of stationarity with a one-time structural break, accounting for regime shifts that bias standard panel unit root tests toward false non-rejection.

Key highlights

  • Endogenously identifies structural break dates — no need to pre-specify break points, avoiding look-ahead bias.
  • Substantially more powerful than standard panel unit root tests (LLC, IPS) when true breaks exist, reducing false acceptance of the unit root null.
  • Allows heterogeneous break dates across cross-sectional units, reflecting country- or firm-specific shocks.
  • Accommodates both level breaks and trend breaks through flexible model specifications (Models A, B, and C).
  • Consistent with the Zivot-Andrews (1992) framework, providing a well-grounded theoretical basis.

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 Panel Zivot-Andrews test when you have panel time-series data and suspect that structural breaks — caused by economic crises, policy reforms, or institutional changes — may distort the verdict of standard panel unit root tests (LLC, IPS, Fisher-ADF). It is appropriate when break dates are unknown and may differ across units. Prefer it over conventional tests when the time dimension is moderate (T ≥ 20 per unit) and N is manageable. Do not use it when the time dimension is very short (T < 15), when breaks are a priori impossible, or when you need a stationarity (rather than unit root) null — use KPSS-based tests in that case.

Strengths & limitations

Strengths
  • Endogenously identifies structural break dates — no need to pre-specify break points, avoiding look-ahead bias.
  • Substantially more powerful than standard panel unit root tests (LLC, IPS) when true breaks exist, reducing false acceptance of the unit root null.
  • Allows heterogeneous break dates across cross-sectional units, reflecting country- or firm-specific shocks.
  • Accommodates both level breaks and trend breaks through flexible model specifications (Models A, B, and C).
  • Consistent with the Zivot-Andrews (1992) framework, providing a well-grounded theoretical basis.
Limitations
  • Assumes at most one structural break per unit; multiple breaks require extensions such as Lumsdaine-Papell or Bai-Perron approaches.
  • Power degrades when T is small (T < 20 per unit), as the break-search procedure consumes degrees of freedom.
  • Critical values depend on the assumed model type (level break, trend break, or both); misspecifying the model inflates size distortion.
  • Combining heterogeneous unit tests into a single panel statistic requires assumptions about cross-sectional independence that may be violated.

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 does the Panel Zivot-Andrews test differ from the standard Panel ADF test?

The Panel ADF (Fisher-ADF, LLC, IPS) assumes no structural break in the data-generating process. If a true break exists, standard tests are biased toward non-rejection of the unit root. The Panel Zivot-Andrews test explicitly models a one-time break for each unit, yielding higher power under the stationary-with-break alternative.

What is the difference between Model A, B, and C in the Zivot-Andrews framework?

Model A allows only a level (intercept) break; Model B allows only a slope (trend) break; Model C allows both a level and a slope break. Model C is the most general and is recommended when there is no strong prior about the break type, though it requires larger samples to maintain adequate power.

Can the Panel Zivot-Andrews test handle more than one structural break per unit?

No — it is designed for a single break per unit. For series with multiple breaks, use the Lumsdaine-Papell (two-break) test or the Bai-Perron multiple breakpoint framework at the unit level, then combine across the panel.

What do I do if my panel has cross-sectional dependence?

Standard combination rules (Fisher, IPS) assume cross-sectional independence. With cross-sectional dependence, use bootstrap-based critical values or apply the CIPS-type corrections introduced by Pesaran (2007) adapted to the structural break setting.

What minimum time dimension T is recommended?

Practitioners generally recommend T ≥ 20 per unit. Shorter series reduce the number of valid break-search candidates and inflate size distortions. With T < 15, standard Panel ADF with robust standard errors is usually more reliable.

Sources

  1. 1.
    Zivot, E., & Andrews, D. W. K. (1992). Further evidence on the great crash, the oil-price shock, and the unit-root hypothesis. Journal of Business & Economic Statistics, 10(3), 251–270.
    DOI 10.1080/07350015.1992.10509904
  2. 2.
    Pedroni, P. (1999). Critical values for cointegration tests in heterogeneous panels with multiple regressors. Oxford Bulletin of Economics and Statistics, 61(S1), 653–670.
    DOI 10.1111/1468-0084.0610s1653

You have read it. What now?

Ask about this methodSimilar methods

Cite this page

ScholarGate. (2026, June 3). Panel Zivot-Andrews test. ScholarGate. https://scholargate.app/econometrics/panel-zivot-andrews-test