BEKK-GARCH: Multivariate Conditional Volatility Modeling
BEKK Multivariate GARCH · Also known as: BEKK Model, Baba-Engle-Kraft-Kroner GARCH, Multivariate BEKK, BEKK-ÇARCH Modeli
BEKK-GARCH, proposed by Engle and Kroner (1995), is a multivariate GARCH specification that models the time-varying conditional covariance matrix of a system of financial return series. Named after Baba, Engle, Kraft, and Kroner, it is the dominant framework for quantifying volatility spillovers and dynamic correlations across multiple assets or markets simultaneously, widely adopted by financial economists and risk managers since the mid-1990s.
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
BEKK-GARCH is appropriate when researchers need to model time-varying covariances among two to five financial return series simultaneously and wish to test for volatility spillovers or dynamic hedging ratios. Key assumptions include covariance stationarity of the return process and that conditional second moments follow the BEKK recursion. It is best suited for daily or higher-frequency financial data. For systems with more than five assets the parameter count grows rapidly, making DCC-GARCH or factor GARCH preferable. BEKK is not suited for non-financial count or proportion data.
Strengths & limitations
- Guarantees positive definiteness of the conditional covariance matrix without auxiliary constraints.
- Directly models cross-asset shock and volatility transmission through off-diagonal parameter matrices.
- Quasi-maximum likelihood estimation is consistent and asymptotically normal under mild regularity conditions.
- Enables derivation of time-varying optimal hedge ratios and portfolio weights from the estimated covariance path.
- Parameter proliferation: a full BEKK(1,1) model for N assets requires O(N^2) parameters, making estimation impractical beyond roughly five series.
- Computational burden is high; numerical optimization of the log-likelihood can be slow and sensitive to starting values.
- Does not separately identify individual shock and variance components as cleanly as scalar or diagonal BEKK variants.
- Interpretation of individual parameter matrices A and G is not straightforward without additional structural assumptions.
Frequently asked
What does the BEKK acronym stand for?
BEKK stands for the initials of the four authors—Baba, Engle, Kraft, and Kroner—who developed the model in a 1990 working paper. When Engle and Kroner formally published the model in 1995, they retained the four-author acronym to credit all contributors to the original manuscript.
How does BEKK-GARCH differ from DCC-GARCH?
BEKK models the full conditional covariance matrix jointly through a quadratic recursion, enabling direct estimation of cross-asset shock transmission but at the cost of many parameters. DCC-GARCH separates the problem into univariate volatilities and a dynamic correlation component, scaling better to large systems but imposing a more restrictive structure on how correlations evolve over time.
When should the diagonal BEKK restriction be preferred over the full BEKK?
Diagonal BEKK restricts matrices A and G to be diagonal, eliminating cross-asset shock transmission in the variance equation and drastically reducing the parameter count from O(N^2) to O(N). It is preferred when the primary interest is in own-variance dynamics rather than spillovers, or when the sample size is insufficient to reliably identify the full off-diagonal parameter set.
Sources
- Engle, R. F., & Kroner, K. F. (1995). Multivariate simultaneous generalized ARCH. Econometric Theory, 11(1), 122–150. DOI: 10.1017/S0266466600009063 ↗
How to cite this page
ScholarGate. (2026, June 2). BEKK Multivariate GARCH. ScholarGate. https://scholargate.app/en/econometrics/bekk-garch
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 →