Which is Worse: Heavy Tails or Volatility Clusters?

Abstract

Heavy tails and volatility clusters are stylized facts of financial returns that destabilize markets but are often overlooked by assuming normally distributed or iid returns. This work disentangles the two facts and is the first to assess which one does the greater damage to financial stability and whether diversification can reduce the threat. Thereby, it also quantifies the potential shortfalls of the two simplifying assumptions. The analysis covers seven index return series representing different asset classes and individual stock portfolios. The stylized facts are isolated using a novel modeling approach that leverages recent developments in surrogate analysis (IAAFT, IAAWT), which outperforms common methods like GARCH filters or stochastic process simulations. Our results shows that volatility clusters significantly impact maximum drawdowns and aggregate losses across all markets and that diversification does not yield protection against those risks. In fact, it amplifies the translation of the two stylized facts into drawdowns. We further demonstrate the practical relevance of our findings as we replicate our surrogate analysis findings with real portfolios and show that regulators should consider the impact of volatility clusters and abandon the iid assumption to improve the accuracy of capital buffers.

Keywords: Financial Stability, Tail Risk, Autocorrelation, Volatility Clustering, Heavy Tails, Risk Management

JEL Classification: G12, G18, G15, G01