Fourier ARDL Bounds Test
Fourier Autoregressive Distributed Lag Bounds Test · Also known as: Fourier ARDL, Fourier bounds testing, ARDL with Fourier approximation, F-ARDL cointegration test
The Fourier ARDL bounds test augments the Pesaran-Shin-Smith cointegration framework with trigonometric (Fourier) terms that capture gradual, smooth structural breaks in the data-generating process. It tests for a long-run level relationship between variables without requiring the researcher to specify the number, timing, or form of structural breaks in advance.
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.
+10 more
When to use it
Use the Fourier ARDL bounds test when you suspect smooth, gradual structural change in the relationship between variables — common in long macroeconomic or financial panel series — and you cannot or do not wish to impose specific break dates. It is appropriate for small-to-moderate samples, handles variables of mixed integration order I(0)/I(1), and is robust to abrupt-break misspecification as long as breaks are gradual. Do not use it when structural breaks are genuinely abrupt and sharp (prefer Zivot-Andrews or Bai-Perron alternatives), when the sample is very short (fewer than 40-50 observations), or when regressors are I(2), which invalidates the bounds critical values.
Strengths & limitations
- Does not require the researcher to pre-specify the number, timing, or form of structural breaks.
- Handles mixed-order regressors (I(0) and I(1)) without pre-testing for unit roots in each variable.
- More powerful than standard ARDL when structural breaks are smooth and gradual rather than abrupt.
- The Fourier approximation is parsimonious — one or two frequency components often suffice.
- Retains the simple OLS/NLS estimation framework and standard bounds critical values.
- If true breaks are sharp and discontinuous, the Fourier approximation may fit poorly and reduce power.
- Requires a sufficient time-series length (generally T > 40-50) for the Fourier terms to be estimated reliably.
- Critical value tables must be adjusted for the Fourier augmentation; applying standard Pesaran et al. tables without modification can distort size.
- Selecting the Fourier frequency k by grid search can overfit in short samples if K is set too large.
Frequently asked
How is the Fourier ARDL bounds test different from the standard ARDL bounds test?
The standard ARDL bounds test assumes a constant intercept and trend over the whole sample. The Fourier version adds sine-cosine terms that let the deterministic component change smoothly, making the test more robust when gradual structural shifts are present. Without Fourier terms, the standard test may fail to detect cointegration that exists only in certain sub-periods.
How do I choose the Fourier frequency k?
Estimate the model for each integer k from 1 to some maximum K (typically K = 5 or T/2, whichever is smaller) and select the k that minimises the residual sum of squares or AIC. Only a single frequency is usually retained to keep the model parsimonious.
Can I use this test if I am unsure whether my variables are I(0) or I(1)?
Yes — one key advantage inherited from the ARDL framework is that the bounds test is valid for regressors of mixed integration order. Pre-testing for unit roots is informative but not strictly required, provided no variable is I(2).
What critical values should I use?
You need critical value bounds that account for the Fourier augmentation, as reported in papers specifically developing the Fourier ARDL test. Using the original Pesaran et al. (2001) tables without modification may over-reject the null of no cointegration.
What is the minimum sample size for the Fourier ARDL bounds test?
There is no universally fixed minimum, but simulation studies suggest the test has acceptable size and power at T = 50 or more. Very short samples (T < 40) make the Fourier terms difficult to estimate precisely and weaken inference.
Sources
- Nazlioglu, S., Gormus, A., & Soytas, U. (2021). Oil prices and monetary policy in emerging markets: structural breaks, asymmetries, and Fourier approximations. Energy Economics, 95, 105119. link ↗
- Pesaran, M. H., Shin, Y., & Smith, R. J. (2001). Bounds testing approaches to the analysis of level relationships. Journal of Applied Econometrics, 16(3), 289-326. DOI: 10.1002/jae.616 ↗
How to cite this page
ScholarGate. (2026, June 3). Fourier Autoregressive Distributed Lag Bounds Test. ScholarGate. https://scholargate.app/en/econometrics/fourier-ardl-bounds-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.
- ARDL Bounds TestEconometrics↔ compare
- Fourier Engle-Granger cointegrationEconometrics↔ compare
- Nonlinear ARDLEconometrics↔ compare
- Structural Break ARDL Bounds TestEconometrics↔ compare
- Vector Error Correction ModelEconometrics↔ compare