Professor of Finance, Dalhousie University
Canada Research Chair (tier II) in Risk Management
The Internet bubble and 2008-2009 economic crash exposed severe limitations of traditional portfolio models, especially the dependence on a static framework e. g. a constant covariance matrix. This paper develops a novel dynamic optimization model for constructing a long-short equity portfolio. A hidden Markov model captures the critical market sentiments, with expected asset returns highly dependent on the associated economic regimes. Expected equity returns are characterized by a set of eight economic factors within a regime-switching auto-regressive approach. In the empirical analysis, we employ exchange traded funds to test the approach. Common factors include: changes in the S&P 500 price index, Treasury bond index, the U.S. dollar index, implied volatility, aggregate dividend yield, short term interest rate, treasury yield spread, and credit spread. The optimal portfolio is subject to the various policy constraints on leverage, individual positions, and Value at Risk. The portfolio exposure to each of the risk factors is controlled by the level of risk aversion. The empirical tests show that the developed investment portfolio provides much higher returns with very limited risk, in contrast with alternative investment approaches, for the period of January 1999 to November 2010.