A0565
Title: High-dimensional sparse factor multivariate stochastic volatility models
Authors: Manabu Asai - Soka University (Japan) [presenting]
Benjamin Poignard - Osaka University (Japan)
Abstract: Factor models are useful for reducing the dimension of variables. Factor structure is accommodated on high-dimensional multivariate stochastic volatility (MSV) models to construct the space factor MSV (fMSV) model. Two sparse factor models are considered; one assumes sparsity for the covariance matrix of the idiosyncratic errors, while the other has sparsity on the factor loading matrix. Some theoretical results are provided for the fMSV models. Using simulated and real data, the in-sample and out-of-sample forecasting performance is examined, comparing the fMSV models with DCC and BEKK models.
