Fourier Phillips-Perron (Fourier PP) Unit Root Test
Also known as: Fourier PP test, Flexible Fourier PP unit root test, Enders-Lee Fourier PP test, nonlinear PP unit root test
The Fourier PP unit root test extends the classical Phillips-Perron test by embedding low-frequency Fourier terms in the deterministic component, enabling the test to account for an unknown number of smooth, gradual structural breaks in the level or trend without pre-specifying their timing or shape.
Key highlights
- Accounts for smooth structural breaks without requiring the researcher to specify their number, timing, or functional form.
- Inherits the non-parametric serial-correlation robustness of the Phillips-Perron correction, avoiding lag-selection sensitivity.
- More powerful than standard unit root tests when the true DGP has gradual breaks, reducing the risk of false non-rejection of the unit root null.
- A single parsimonious Fourier component can approximate a wide variety of smooth nonlinear trend shapes.
- Applicable to a broad class of economic and financial series where gradual regime changes are common.
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 Fourier PP test when you suspect a time series may be stationary around a smoothly changing trend or level — for example, after economic liberalisation, a gradual policy shift, or a prolonged business cycle. It is preferable to standard PP when smooth, unknown structural breaks are plausible but their number and timing are not known in advance. Do not use it when breaks are sharp and abrupt (prefer Zivot-Andrews or Lee-Strazicich in that case), when the sample is very short (fewer than ~80 observations), or as a mechanical substitute for the standard PP without first checking whether Fourier terms significantly improve the fit.
Strengths & limitations
- Accounts for smooth structural breaks without requiring the researcher to specify their number, timing, or functional form.
- Inherits the non-parametric serial-correlation robustness of the Phillips-Perron correction, avoiding lag-selection sensitivity.
- More powerful than standard unit root tests when the true DGP has gradual breaks, reducing the risk of false non-rejection of the unit root null.
- A single parsimonious Fourier component can approximate a wide variety of smooth nonlinear trend shapes.
- Applicable to a broad class of economic and financial series where gradual regime changes are common.
- Simulation-based critical values depend on the sample size and deterministic specification; standard PP tables cannot be used.
- Low power in small samples (T < 80), especially when the true break is small or occurs near the sample endpoints.
- Does not perform well against sharp, sudden breaks — in that setting, Zivot-Andrews or Lee-Strazicich tests are more appropriate.
- Fourier approximation may overfit or underfit if the number of candidate frequencies is not chosen carefully.
- The test does not identify the location or magnitude of structural breaks; supplementary break-dating methods are needed for interpretation.
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 Fourier PP test differ from the standard Phillips-Perron test?
The standard PP test assumes a constant or linear deterministic trend. The Fourier PP test augments this with sine and cosine terms that approximate smooth, nonlinear trend changes, making it robust to gradual structural breaks. Critical values differ and must come from simulation.
How do I choose the Fourier frequency k?
The optimal frequency k is chosen by searching over a small grid (typically k = 1 to 5) and selecting the value that minimises the residual sum of squares from the Fourier regression. Lower frequencies capture broader, slower shifts.
Should I always prefer the Fourier PP test over the standard PP test?
No. If the Fourier sine and cosine terms are jointly insignificant (tested with an F-test), the smooth-break component is unnecessary and the standard PP test is more parsimonious. Only use the Fourier extension when smooth trend variation is substantively plausible.
What should I do if the Fourier PP test and the standard PP test give contradictory conclusions?
Report both results and consider the economic context. Divergence often indicates the presence of smooth breaks: the standard PP spuriously rejects or fails to reject the null because it cannot separate trend variation from the unit root component.
Can the Fourier PP test be used with panel data?
Yes, panel extensions of the Fourier unit root testing approach exist (see Panel Fourier PP), but they require different critical values and panel-specific implementation. The univariate Fourier PP applies to a single time series.
Sources
- 1.Enders, W., & Siklos, P. L. (2001). Cointegration and threshold adjustment. Journal of Business and Economic Statistics, 19(2), 166-176.
- 2.Becker, R., Enders, W., & Lee, J. (2006). A stationarity test in the presence of an unknown number of smooth breaks. Journal of Time Series Analysis, 27(3), 381-409.
You have read it. What now?
Cite this page
ScholarGate. (2026, June 3). Fourier PP unit root test. ScholarGate. https://scholargate.app/econometrics/fourier-pp-unit-root-test