July 1, 2002Journal of Business and Economic Statistics
Dynamic Conditional Correlation
Authors
Discussion
Loading...
Member takes
Implication
Methodological evaluation demonstrates dynamic conditional correlation modeling in multivariate time series, highlighting an efficient framework for time-varying market covariance.
Key Points
- To develop a flexible and computationally parsimonious multivariate framework for estimating time-varying conditional correlations across dynamic systems.
- Formulated dynamic conditional correlation models combining univariate GARCH volatility structures with parametric specifications for correlation dynamics.
- Designed an estimation approach utilizing univariate and two-step methods based on the joint likelihood function.
- Dynamic conditional correlation models successfully captured nonlinear dependency structures without the computational burden typical of conventional multivariate GARCH models.
- The parametric modeling structure delivered robust, empirically sensible estimates of time-varying correlation across diverse test conditions.
Cite This Study
Robert F. Engle (2002) studied this question.