This research studies asset allocation in the presence of regime switching in asset returns, with two underlying models, the Multivariate Regime Switching Model and the Univariate Regime Switching Model. To our knowledge this thesis is the first to construct and compare the Univariate Regime Switching Model with the Multivariate Regime Switching Model both statistically as distribution parameters estimators and with respect to their application to the asset allocation framework. The asset allocation decision in this thesis is based on the two before mentioned Regime Switching Models and compared to the more familiar case of a non-regime switching model that relies on the mean-variance framework where asset returns are assumed to be normal. Allowing the returns distribution to follow a joint distribution of two regimes, characterized as bear and bull markets, allows us to make more realistic approximations and simulations of real financial markets, which furthermore, allows the asset allocation strategies to take into account the so called extreme events, namely the fat tails of the returns distribution. The first part of this research is devoted to analyzing the raw data, determining regimes and estimating the parameters corresponding to those regimes which are found to be significant. The regimes identified are named bear regime, characterized with high volatility in returns and negative mean returns, and bull regime, characterized with low volatility and positive mean returns. Furthermore, as two different Regime Switching Models are constructed, a statistical comparison is carried out. The second part of the research makes use of the two different identifications of regimes in asset allocation models. The two asset allocation models are compared individually as well as with a non-switching model. All models exhibit different results for the asset allocation decision, both with respect to the length of the investment horizon and to various levels of risk aversion. Furthermore we both consider the asset allocation case of a buy-and-hold strategy as well as allowing for rebalancing of the portfolio. Overall this research shows that using Regime Switching Models allows us to better represent the current financial markets, as we are able to go beyond the unsatisfactory assumption of the normal distribution of asset returns. This allows us to better model the turbulent financial markets by taking into consideration fat tails in returns distributions when implementing asset allocation strategies.
|Educations||MSc in Applied Economics and Finance, (Graduate Programme) Final Thesis|
|Number of pages||114|