Nonlinear EGARCH Model
Nonlinear Exponential Generalized Autoregressive Conditional Heteroscedasticity Model · Also known as: NL-EGARCH, nonlinear exponential GARCH, asymmetric EGARCH, NEGARCH
The Nonlinear EGARCH model extends Nelson's (1991) Exponential GARCH by allowing the news impact function to take a flexible nonlinear form, capturing asymmetric and nonlinear responses of conditional volatility to past shocks. It is widely used in financial econometrics to model leverage effects and complex volatility dynamics in asset returns.
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 the Nonlinear EGARCH model when you have high-frequency or daily financial return series that exhibit volatility clustering, strong leverage effects, and there is evidence that the news impact curve is not well captured by a standard GARCH or linear EGARCH — for instance when the Engle-Ng sign-bias test flags nonlinearity. It is appropriate for equity returns, exchange rates, and commodity prices where tail risk and asymmetric volatility matter. Do not use it when the series shows no significant ARCH effects (Engle's LM test), when the sample is small (fewer than ~500 observations), or when interpretability of a parsimonious model outweighs a flexible but harder-to-interpret nonlinear fit.
Strengths & limitations
- Models volatility in log-space, guaranteeing positive conditional variance without sign constraints on parameters.
- Captures the leverage effect: negative shocks can raise volatility proportionally more than positive shocks of the same magnitude.
- Nonlinear news impact function allows flexible, data-driven volatility responses beyond the fixed asymmetric shape of standard EGARCH.
- Heavy-tailed error distributions (Student-t, GED) are straightforward to incorporate via maximum likelihood.
- Impulse-response analysis of volatility shocks is more intuitive in the log-variance parameterization.
- Nonlinear extensions introduce additional parameters, increasing estimation complexity and the risk of overfitting in small samples.
- Maximum likelihood estimation can be sensitive to starting values and may converge to local optima.
- Interpretation of nonlinear news impact parameters is less transparent than standard EGARCH or GJR-GARCH.
- Forecasting beyond short horizons is difficult because nonlinear volatility dynamics do not aggregate simply.
- Requires large samples (typically 500+ daily observations) for reliable estimation of the nonlinear component.
Frequently asked
How does the Nonlinear EGARCH differ from the standard EGARCH?
Standard EGARCH uses a fixed linear-asymmetric news impact function g(z_t) = θz_t + γ(|z_t| − E|z_t|). The nonlinear variant replaces or augments this with a flexible functional form — such as a smooth transition or polynomial — allowing the volatility response to shocks to vary across different shock sizes or market regimes, providing a better fit when the standard shape is rejected by diagnostic tests.
Why model log(σ²) rather than σ² directly as in GARCH?
Modelling the log of conditional variance ensures σ²_t is always positive regardless of the sign of the estimated parameters. In standard GARCH, non-negativity constraints must be imposed explicitly, complicating estimation. The log parameterization also makes the model more stable and interpretable in terms of percentage changes in volatility.
How do I test whether the nonlinear extension is needed?
The Engle-Ng (1993) sign-bias test decomposes the squared standardized residuals of a baseline GARCH or EGARCH model into components driven by the sign and size of past shocks. Significant joint test statistics indicate that the linear asymmetric form is inadequate and a nonlinear specification should be considered.
What sample size is needed for reliable estimation?
As a rule of thumb, at least 500 daily observations are recommended for a standard EGARCH; nonlinear extensions with additional parameters may require 1,000 or more to avoid overfitting and to achieve numerical stability in maximum likelihood estimation.
Can the Nonlinear EGARCH be combined with multivariate models?
Yes. Univariate Nonlinear EGARCH models can serve as the marginal volatility specifications within multivariate frameworks such as DCC (Dynamic Conditional Correlation), providing flexible volatility inputs for portfolio risk and correlation modelling.
Sources
- Nelson, D. B. (1991). Conditional heteroskedasticity in asset returns: A new approach. Econometrica, 59(2), 347–370. DOI: 10.2307/2938260 ↗
- Engle, R. F., & Ng, V. K. (1993). Measuring and testing the impact of news on volatility. Journal of Finance, 48(5), 1749–1778. DOI: 10.1111/j.1540-6261.1993.tb05127.x ↗
How to cite this page
ScholarGate. (2026, June 3). Nonlinear Exponential Generalized Autoregressive Conditional Heteroscedasticity Model. ScholarGate. https://scholargate.app/en/econometrics/nonlinear-egarch-model
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.
- ARCH modelEconometrics↔ compare
- EGARCH modelEconometrics↔ compare
- GARCH ModelEconometrics↔ compare
- Stochastic Volatility ModelFinance↔ compare
- TGARCH modelEconometrics↔ compare