Multivariate Financial Volatility Modelling

Volatility and correlation of asset prices are crucial for many finance applications and policy tasks. Therefore, this lecture treats multivariate models for variances and covariances of financial time series. In detail, we mainly deal with variants of autoregressive conditional heteroscedasticity (ARCH). The VECH and BEKK models are introduced as straightforward multivariate extensions of univariate ARCH processes. In this context, we discuss the problems of high dimensionality and positive definiteness of the covariance matrix. Solutions are offered by constant and dynamic conditional correlation models. Additionally, the lecture covers heteroscedastic factors and latent variables structures, which exhibit both econometric appeal and a direct link to finance theory. In the same line, stochastic volatility (SV) is presented as a process that treats volatility itself as a latent process. For estimation in presence of such non-observable variables, Kalman filtering is employed. At last, we touch on the concept of realised (co-)volatility that exploits high-frequency data for improving variance measurement and modelling. The discussed approaches are applied in computer tutorials using the Gauss system.

Voraussetzungen

Methoden der ?konometrie; Applied Financial Econometrics empfehlenswert

Downloads: Grips (Pa?wort wird in 1. VL bekanntgegeben.)

Gliederung

Kontakt

Enzo Weber

Turnus

unregelm??ig, derzeit nicht angeboten