Structural Break ADF Unit Root Test
Structural Break Augmented Dickey-Fuller Unit Root Test · Also known as: ADF with structural break, Perron unit root test, break-augmented ADF, unit root test with structural change
The structural break ADF unit root test extends the standard Augmented Dickey-Fuller test to allow for one or more discrete shifts in the level or trend of a time series. Because ignoring a structural break inflates the apparent persistence of a series, this test prevents false acceptance of the unit root null when the series is actually stationary around a shifting mean or trend.
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 this test when you have a single time series that may have experienced a discrete structural change — caused by a policy reform, a major economic event, or a crisis — and you want to test for a unit root while accounting for that break. It is preferable to the standard ADF whenever a visual inspection, a Chow test, or prior economic knowledge suggests a break in mean or trend. Do not use it when you suspect multiple breaks (use Bai-Perron procedures instead), when the series is very short (fewer than 50 observations), or when you have no theoretical justification for a break — searching for breaks purely data-drivenly over many series inflates size.
Strengths & limitations
- Prevents the well-documented size distortion of the standard ADF test caused by ignored structural breaks.
- The Zivot-Andrews variant does not require prior knowledge of the break date, making it applicable in practice.
- Distinguishes between a unit root and stationarity around a shifting level or trend, which has important implications for long-run forecasting and cointegration analysis.
- Produces an endogenously estimated break date as a by-product, which can be economically meaningful.
- Non-standard but well-tabulated critical values are available for the most common break specifications.
- Designed for a single structural break; multiple breaks require more general procedures such as Bai-Perron.
- Critical values depend on the assumed break form (intercept only, trend only, or both) and are not interchangeable.
- Low power in short samples or when the break occurs near the boundaries of the sample.
- The endogenous break-date search in Zivot-Andrews inflates the probability of detecting a spurious break if the series truly has a unit root.
Frequently asked
How does this test differ from the standard ADF test?
The standard ADF test assumes a fixed, unbroken deterministic component (constant and/or trend). The structural break ADF adds a dummy variable to capture a one-time shift in level or trend, preventing the break from inflating the estimated degree of persistence and biasing the test toward falsely accepting the unit root null.
How do I choose whether the break affects the intercept, the trend, or both?
Economic theory and visual inspection of the series should guide the choice. A sudden level shift (e.g., a one-off policy change) suggests an intercept break; a change in the growth rate suggests a trend-slope break; a major regime change may justify both. Running all three specifications and checking consistency of conclusions is a common practice.
What if I have more than one break?
The single-break ADF has low power against multiple breaks and may give misleading results. For two or more suspected breaks, use the Bai-Perron (1998, 2003) multiple-break framework or a unit root test that accommodates multiple breaks, such as those of Lee and Strazicich (2003).
Do the standard ADF critical values apply here?
No. Because the break dummy changes the asymptotic distribution, critical values are non-standard. Perron (1989) and Zivot-Andrews (1992) tabulate the relevant critical values for each break specification; most econometric software implements them automatically.
Can I use this test if I suspect the break date but am not certain?
Yes — use the Zivot-Andrews endogenous variant, which searches over all interior break dates and selects the one that gives the smallest (most negative) t-statistic. This removes the need for a priori knowledge of the exact break date while remaining valid under the null.
Sources
- Perron, P. (1989). The great crash, the oil price shock, and the unit root hypothesis. Econometrica, 57(6), 1361-1401. DOI: 10.2307/1913712 ↗
- 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 and Economic Statistics, 10(3), 251-270. DOI: 10.1080/07350015.1992.10509904 ↗
How to cite this page
ScholarGate. (2026, June 3). Structural Break Augmented Dickey-Fuller Unit Root Test. ScholarGate. https://scholargate.app/en/econometrics/structural-break-adf-unit-root-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.
- Augmented Dickey-Fuller unit root testEconometrics↔ compare
- Phillips-Perron unit root testEconometrics↔ compare
- Structural Break Granger CausalityEconometrics↔ compare
- Structural Break KPSS TestEconometrics↔ compare
- Structural break PP unit root testEconometrics↔ compare
- Zivot-Andrews Structural Break TestEconometrics↔ compare