Panel TGARCH (Threshold GARCH for Panel Data)
Also known as: Panel GJR-GARCH, Panel Asymmetric GARCH, Panel Threshold GARCH, TGARCH panel model
Panel TGARCH extends the Threshold GARCH (GJR-GARCH) model to panel data, allowing each cross-sectional unit to exhibit asymmetric volatility responses — where negative shocks generate larger variance increases than positive shocks of the same magnitude — while exploiting the cross-sectional dimension to obtain more efficient parameter estimates.
Key highlights
- Captures the leverage effect — the asymmetric impact of negative versus positive shocks on conditional variance — within a rigorous econometric framework.
- Pools information across cross-sectional units, yielding more efficient estimates than fitting separate TGARCH models for each unit.
- Accommodates unit-specific baseline volatility through fixed or random effects, preserving heterogeneity.
- QML estimation is consistent even when innovations are non-Gaussian, making it robust to fat-tailed distributions common in financial data.
- The threshold parameter γ provides a direct, testable measure of asymmetry that can be compared across panels.
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 Panel TGARCH when you have panel data (e.g., multiple stocks, firms, or countries observed over time) and theory or prior evidence suggests asymmetric volatility — specifically that negative shocks raise variance more than positive shocks of equal size. It is particularly appropriate in empirical asset pricing, risk management, and macroeconomic volatility studies where pooling cross-sections improves efficiency but unit heterogeneity must be respected. Avoid it when the time dimension is very short (T < 30 per unit), when the data generating process is better described by symmetric GARCH or EGARCH, or when cross-sectional dependence (common factors) is strong and unaccounted for, as this can invalidate standard errors.
Strengths & limitations
- Captures the leverage effect — the asymmetric impact of negative versus positive shocks on conditional variance — within a rigorous econometric framework.
- Pools information across cross-sectional units, yielding more efficient estimates than fitting separate TGARCH models for each unit.
- Accommodates unit-specific baseline volatility through fixed or random effects, preserving heterogeneity.
- QML estimation is consistent even when innovations are non-Gaussian, making it robust to fat-tailed distributions common in financial data.
- The threshold parameter γ provides a direct, testable measure of asymmetry that can be compared across panels.
- Requires a sufficiently long time dimension (typically T ≥ 30 per unit) for reliable volatility estimation; short panels lead to imprecise estimates.
- Strong cross-sectional dependence (e.g., common global shocks) can bias inference if not explicitly modeled with cross-section robust standard errors or factor-augmented specifications.
- Estimation is computationally intensive, and convergence can be sensitive to starting values in large panels.
- The fixed-effects panel GARCH framework does not have a clean within-group transformation, so standard panel FE tricks do not directly apply — specialised software or iterative estimators are needed.
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
What distinguishes Panel TGARCH from Panel GARCH?
Panel GARCH assumes symmetric volatility responses to positive and negative shocks. Panel TGARCH adds a threshold term that allows negative shocks to have a larger impact on conditional variance, capturing the leverage effect. If the asymmetry parameter γ is not statistically significant, Panel TGARCH reduces to Panel GARCH.
How do I test whether the leverage effect is significant?
Estimate the model and conduct a standard t-test (or Wald test) on γ. A significantly positive γ confirms that negative innovations inflate future volatility more than positive ones of equal magnitude. Some software also reports a likelihood ratio test comparing the symmetric Panel GARCH to the Panel TGARCH specification.
Should I use fixed effects or random effects for the intercept ω_i?
Run a Hausman-type test. If unit-specific intercepts are correlated with the regressors in the mean equation, fixed effects are consistent; otherwise random effects are efficient. In practice, the panel GARCH literature often defaults to fixed effects or reports both for robustness.
What software can estimate Panel TGARCH?
R packages such as panelGARCH and rugarch (looped over units with pooled constraints) support variants of panel GARCH and TGARCH. Stata's xtgarch community-contributed package and custom MLE routines in Python (arch library) are also used. Some researchers code bespoke panel TGARCH estimators for specific pooling restrictions.
How do I handle cross-sectional dependence in Panel TGARCH?
First test for cross-sectional dependence using the Pesaran CD test. If dependence is present, use panel-robust (Driscoll-Kraay) standard errors, demean the series by a common factor before estimation, or augment the variance equation with a common volatility factor.
Sources
- 1.Glosten, L. R., Jagannathan, R., & Runkle, D. E. (1993). On the relation between the expected value and the volatility of the nominal excess return on stocks. Journal of Finance, 48(5), 1779–1801.
- 2.Zakoian, J.-M. (1994). Threshold heteroskedastic models. Journal of Economic Dynamics and Control, 18(5), 931–955.
You have read it. What now?
Cite this page
ScholarGate. (2026, June 3). Panel TGARCH. ScholarGate. https://scholargate.app/econometrics/panel-tgarch