Fourier EGARCH: Volatility Modeling with Smooth Structural Breaks
Fourier Exponential Generalized Autoregressive Conditional Heteroscedasticity · Also known as: Fourier-EGARCH, F-EGARCH, Fourier exponential GARCH, smooth structural break EGARCH
Fourier EGARCH extends Nelson's (1991) Exponential GARCH model by embedding Fourier trigonometric terms in the conditional variance equation to capture smooth, gradual shifts in the unconditional variance level over time. This allows the model to handle structural breaks in volatility without requiring prior knowledge of their timing or number.
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 Fourier EGARCH when modeling financial return volatility over a long sample that likely spans multiple market regimes or structural changes, and when you want to avoid specifying break dates ad hoc. It is particularly appropriate for equity, exchange rate, or commodity return series where both volatility clustering and leverage effects are expected and where standard EGARCH residuals show signs of non-stationarity in variance. Do not use it with short samples (fewer than about 500 observations) where Fourier terms are poorly identified, or when volatility breaks are sharp and abrupt rather than gradual — in that case, Markov-switching GARCH or dummy-break EGARCH models are more appropriate.
Strengths & limitations
- Captures smooth, gradual structural breaks in unconditional volatility without specifying break dates in advance.
- Retains the leverage-effect and asymmetry modelling of standard EGARCH.
- Avoids positivity constraints on parameters because the variance equation is in logarithmic form.
- More parsimonious than piecewise-break or regime-switching models when breaks are slow-moving.
- Fourier coefficients are estimated jointly with GARCH parameters, so no pre-testing or sequential break detection is required.
- Poorly identified with short samples; requires at least several hundred observations for Fourier terms to be estimated reliably.
- The number of Fourier frequencies K must be chosen by the researcher; selecting too many risks overfitting the in-sample variance path.
- Assumes breaks are smooth and periodic in nature; sharp regime switches or sudden structural breaks are not well captured.
- Increases the parameter count relative to standard EGARCH, which can complicate convergence of the likelihood optimizer.
Frequently asked
How is Fourier EGARCH different from standard EGARCH?
Standard EGARCH assumes a fixed unconditional variance level over the entire sample. Fourier EGARCH adds trigonometric terms to let that level drift smoothly over time, so the model can accommodate prolonged shifts in market volatility — such as the persistent low-volatility period before the 2008 crisis versus the high-volatility period during it — without requiring the researcher to specify break dates.
How do I choose the number of Fourier frequencies K?
Estimate the model for K = 1, 2, and 3 and select the K that minimises AIC or BIC. In practice K = 1 or K = 2 is sufficient for most financial return series covering one to three decades of data.
Can Fourier EGARCH handle sharp, sudden breaks?
No. Fourier terms approximate smooth, gradual changes. If your series contains abrupt regime switches — for example a sudden peg-removal in an exchange rate — a Markov-switching GARCH or a dummy-break specification is more appropriate.
What sample size is needed?
As a practical guideline, at least 500 observations are needed for the Fourier terms to be estimated reliably alongside the GARCH parameters. For daily financial data this corresponds to roughly two years; for weekly or monthly data the requirement is considerably longer.
Does Fourier EGARCH still capture leverage effects?
Yes. The leverage parameter from the standard EGARCH equation is retained intact; Fourier terms only modify the intercept path of the log-variance equation, so asymmetric responses of volatility to positive versus negative shocks are still estimated.
Sources
- Enders, W., & Lee, J. (2012). A unit root test using a Fourier series to approximate smooth breaks. Oxford Bulletin of Economics and Statistics, 74(4), 574-599. DOI: 10.1111/j.1468-0084.2011.00662.x ↗
- Nelson, D. B. (1991). Conditional heteroskedasticity in asset returns: A new approach. Econometrica, 59(2), 347-370. DOI: 10.2307/2938260 ↗
How to cite this page
ScholarGate. (2026, June 3). Fourier Exponential Generalized Autoregressive Conditional Heteroscedasticity. ScholarGate. https://scholargate.app/en/econometrics/fourier-egarch
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.
Compare side by side →