A REGIME SWITCHING MULTIFACTOR MODEL FOR THE STOCK AND BOND RETURNS

Abstract

In contrast to the studies of constant or time-varying correlations between stock and bond returns, in this thesis, I explore the regime-dependent correlations between stock and bond returns. Specifically, I start with a comprehensive asset pricing model, i.e., a regime-switching multifactor model, and then investigate the regime-dependent correlations between stock and bond returns. Based on the BIC, the number of regimes in the regime-switching model is optimally determined to be two. For the two regimes, the directions of the regime-dependent correlations appear to be significantly different. Also, the magnitudes of the regime-dependent correlations are substantially larger in these two regimes than the correlation in the single regime.

With my findings in the regime-dependent correlations, I then examine the performance of portfolio strategies. Throughout the in-sample and out-of-sample tests, I find that the two portfolio strategies, regime inferred portfolio and probability implied portfolio, can outperform the benchmark, S&P 500.