Authors: Alexey I. Balaev
Addresses: Subdepartment of Mathematical Economics and Econometrics, Department of Applied Economics, National Research University Higher School of Economics, Shabolovka str. 26, Moscow 119049, Russia
Abstract: In this paper, we consider multivariate models for returns on Russian equities based on normal distribution, t-distribution with scalar degrees of freedom parameter and t-distribution with vector degrees of freedom parameter. Our models capture autocorrelation, volatility clustering, dynamic links among equity returns and their volatilities, as well as the heavy tails of marginal distributions. Multivariate t-distribution with vector degrees of freedom parameter is a recent generalisation of the classic multivariate t-distribution and in the present paper it is applied to Russian financial data for the first time. Using our multivariate models we construct financial portfolios with conditional minimum variance and conditional maximum expectation of return. We compare optimised portfolios according to the risk and benefit of investment and analyse how the results of this comparison depend on the liquidity of equities involved.
Keywords: portfolio construction; Russian stock market; multivariate t-distribution; vector parameter; degrees of freedom; portfolio performance; multivariate modelling; financial returns; Russia; autocorrelation; volatility clustering; dynamic links; equity returns; heavy tails; marginal distributions; liquidity; investment risk; investment benefits.
International Journal of Computational Economics and Econometrics, 2014 Vol.4 No.1/2, pp.32 - 81
Received: 31 Jul 2013
Accepted: 11 Dec 2013
Published online: 08 Apr 2014 *