How has the relationship between safe haven assets and the US stock market changed after the global financial crisis?

In this paper, I investigate how the relationship between safe haven assets and the US stock market has changed since the global financial crisis. To do so, I propose a new concept of bear-market correlation which allows us to see the effectiveness of the safe haven asset as a hedge and portfolio diversification tool. I compute the bear-market correlation of five safe haven assets during two sample periods, pre-financial crisis (1995–2009) and post-financial crisis (2010–2018). My findings are as follows: First, I find that the empirical bear-market correlations are not explained by multivariate normal and t-distributions. Second, I document that the bear-market correlations of both the Japanese yen and gold show notable changes after the crisis. Third, I estimate a multivariate normal mixture, a multivariate t-distribution mixture, and a generalized dynamic conditional correlation model. I document that these mixture models outperform the dynamic conditional correlation model in most cases. Finally, I discuss the economic impact of failing to capture the bear-market correlation for portfolio optimization.