Publisher:
Lund University, School of Economics and Management, Department of Economics, Lund
Abstract:
This paper extends previous tests of the conditional CAPM using different asymmetric and non-diagonal multivariate GARCH-M specifications for eight large national markets and the world market simultaneously. To solve the well-known problems associated with the likelihood functions of multivariate GARCH models, maximization is performed using simulated annealing, a Markov Chain Monte Carlo stochastic optimization method. We find that a model with double asymmetric effects and a time-varying price of world covariance risk supports all tested asset-pricing restrictions and that the previously often employed symmetric diagonal specification is overwhelmingly rejected. The evidence suggests that investors from all countries could expect statistically significant benefits from international diversification but that gains are considerable larger for investors with smaller home markets than for US and Japanese investors.