Events and Meetings of Italian Statistical Society, Advances in Latent Variables - Methods, Models and Applications

Font Size: 
A Multivariate Latent Stochastic Volatility
Andrea Federico Pierini, Antonello Maruotti

Last modified: 2013-06-16

Abstract


A Multivariate Latent Stochastic Volatility Factor Model is introduced for the estimation of volatility and optimal allocation of stocks portfolio in a Markowitz type portfolio. The proposed model considers the intrinsic values, which regards the value theory, as a function of the market index and the previous returns. The put forward model is applied to a subset of 5 banks among the best capitalized banks’ stocks traded on the Italian stock market (BIT) between 1 January 1986 and 31 August 2011. Computational complexities arising in the estimation step are dealt with a Griddy Gibbs sampler. The association structure among time-series is captured via a factor model.

Full Text: PDF