Title: Penalized quasi-maximum likelihood estimation in extended constant conditional correlation GARCH models
Authors: Marco Avarucci - University of Glasgow (United Kingdom) [presenting]
Stephan Smeekes - Maastricht University (Netherlands)
Abstract: In recent years, multivariate GARCH models have become important tools in empirical finance. For instance, asset pricing and risk management crucially depend on the conditional covariance structure of the assets of a portfolio. Modeling volatility spillovers is also crucial to understand the markets connectedness and contagion. The extended conditional correlation (ECCC) GARCH is attractive for its tractability and ease of interpretation; In particular, conditions for the positive definiteness of the conditional variance and for the existence of strictly stationary solutions are simple and explicit However, when the number of assets is large, the quasi-maximum-likelihood (QML) method - arguably the most popular estimation method in the GARCH setting - can be difficult to apply. To handle this kind of problem, regularization techniques are applied. The asymptotic properties of the penalized QMLE are investigated. Special care has to be taken as the zero parameters lie on the boundary of the parameter space, which affects the asymptotic properties of the estimators.