Structural Break ARCH Model
Autoregressive Conditional Heteroscedasticity Model with Structural Breaks · Also known as: ARCH with structural breaks, break-adjusted ARCH, regime-switching ARCH, SB-ARCH
The Structural Break ARCH model extends Engle's (1982) Autoregressive Conditional Heteroscedasticity framework by explicitly accounting for abrupt, permanent shifts in the conditional variance process. Ignoring structural breaks in variance causes ARCH parameters to appear spuriously persistent, so incorporating break dummies or regime-specific parameters yields more accurate volatility estimates and better model fit.
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 Structural Break ARCH model when a time series exhibits volatility clustering and you have reason to believe the variance level or dynamics shifted at one or more points in the sample — for example, around financial crises, regulatory changes, or major macroeconomic events. It is particularly warranted when a standard ARCH or GARCH model produces implausibly high persistence (alpha + beta near 1 in GARCH terms), which Lamoureux and Lastrapes (1990) showed is often a symptom of unmodelled breaks. Do not use it as a first pass without testing for breaks; imposing arbitrary breaks inflates model complexity. It is also less appropriate when volatility dynamics are better characterised by smooth, gradual change rather than abrupt shifts — in such cases, time-varying parameter ARCH or Markov-switching ARCH may be preferable.
Strengths & limitations
- Corrects the well-documented upward bias in ARCH persistence estimates caused by ignoring structural breaks in variance.
- Provides regime-specific volatility profiles that are more informative for risk management and policy analysis than a single-regime model.
- Compatible with standard maximum likelihood estimation; adding break dummies is computationally straightforward.
- Allows formal likelihood ratio testing to assess whether the breaks are statistically significant, maintaining inferential rigour.
- Improves out-of-sample volatility forecasts when the true process contains regime changes.
- Break dates must be identified before or jointly with estimation; incorrect break placement can introduce new misspecification.
- With many breaks the model becomes heavily parameterised, reducing degrees of freedom and increasing overfitting risk.
- The approach assumes breaks are abrupt and permanent; gradual or transitory shifts are better handled by smooth-transition or Markov-switching variants.
- Does not model the probability of future breaks, limiting its use for long-horizon volatility forecasting beyond the estimation window.
Frequently asked
How do I decide how many structural breaks to include?
Use formal methods such as the ICSS algorithm or the Bai-Perron sequential break test to determine the number and location of breaks. Information criteria (AIC, BIC) and likelihood ratio tests comparing models with k and k+1 breaks also help. Avoid choosing the number of breaks solely to improve model fit, as this inflates parameters and reduces forecasting precision.
Why does ignoring structural breaks make ARCH persistence look higher than it really is?
Lamoureux and Lastrapes (1990) showed that a permanent upward shift in variance appears to a single-regime ARCH model as a very long sequence of large shocks, inflating the estimated persistence of the ARCH process. The model 'thinks' the high variance is dying away slowly, when in reality it simply jumped to a new level. Adding break dummies absorbs this level shift and brings persistence estimates back to their true, lower values.
Can I combine structural breaks with GARCH or asymmetric models?
Yes. The same break-dummy augmentation logic applies to GARCH(p,q), EGARCH, and TGARCH models. Structural break GARCH is the most common extension and is implemented in software such as Eviews, R (strucchange, rugarch packages), and Stata.
What is the difference between a structural break ARCH model and a Markov-switching ARCH model?
The structural break ARCH model assumes breaks are known (or estimated) and permanent, with a deterministic shift at fixed dates. Markov-switching ARCH models treat regimes as latent states that the series can enter and exit probabilistically, capturing recurrent volatility regimes rather than one-time permanent shifts. Choose the structural break approach when breaks are historically identifiable and unlikely to recur in the same form.
Is the structural break ARCH model hard to estimate in practice?
Not significantly harder than standard ARCH. Once break dates are identified, break dummies are added to the variance equation and the model is estimated by standard maximum likelihood. Most econometric packages (Eviews, R rugarch/strucchange, Python arch) support this workflow with minor modifications to the baseline ARCH specification.
Sources
- Engle, R. F. (1982). Autoregressive conditional heteroscedasticity with estimates of the variance of United Kingdom inflation. Econometrica, 50(4), 987–1007. DOI: 10.2307/1912773 ↗
- Lamoureux, C. G., & Lastrapes, W. D. (1990). Persistence in variance, structural change, and the GARCH model. Journal of Business and Economic Statistics, 8(2), 225–234. DOI: 10.1080/07350015.1990.10509794 ↗
How to cite this page
ScholarGate. (2026, June 3). Autoregressive Conditional Heteroscedasticity Model with Structural Breaks. ScholarGate. https://scholargate.app/en/econometrics/structural-break-arch-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
- TGARCH modelEconometrics↔ compare
- Zivot-Andrews Structural Break TestEconometrics↔ compare