The dynamic correlation between stock and government bond returns : empirical evidence from the u.s. market using dcc-garch

The purpose of this master’s thesis is to understand the time-variation in the correlations between U.S. stock and government bond returns. In particular, an underlying objective of this empirical research is to comprehend the behavior of the correlation during financial stress periods. The model that is used to estimate these correlations is the Multivariate DCC-GARCH model as first introduced by Engle in 2001. Eighteen macro-finance factors are selected from the existing literature. Statistical significance of these factors’ impact on the stock-bond correlation is tested through three proposed lagged dependent variable models. The out-of-sample forecasting performance of these models is compared against a benchmark, i.e. the random walk model. The empirical analysis is conducted at both daily and monthly frequencies for a period ranging from January 1997 until November 2014. The empirical results show that high stock volatility explains the flight-to-safety phenomenon during financial stress epochs. Moreover, the findings indicate that the short rate drives the time-variation at daily frequency and that the yield spread and stock liquidity explain the monthly dynamics in stock and government bond return correlations. This research is motivated by its crucial implications on risk management and portfolio optimization.