Dependence between Extreme Events in the Real and Financial Sectors

Extreme events affect both the real economy and financial markets, and it is valuable to understand their interrelationship. We analyze the likelihood of crises in the macroeconomy and in financial markets. We compare rare disaster data from Barro and Jin (2011), crisis data from Reinhart and Rogoff (2009), real time macroeconomic data from Diebold et al (2009), and a unique industry dataset of Turbulence Indices. We examine dependence across the various measures of crises, as well as predictability, using annual and daily data. For annual data, the dependence between crises in the real and financial sectors increases over time for emerging markets, but decreases for OECD countries. We also document persistence at the one year and two year horizons for disasters in the real economy. For daily data, there is, surprisingly, little relation between turbulence in US equity and the real economy. However, there is strong evidence of two-way predictability up to two weeks out for the US economy and global turbulence. A dynamic copula model indicates that the real economy and various turbulence indices alternate between regimes of positive and negative dependence.