High-Frequency Causality in the VIX Index and its derivatives: Empirical Evidence

TL;DR

This paper investigates high-frequency causality between the VIX index and its derivatives during a period of extreme market movement, revealing bidirectional feedback and mean-reversion effects using intraday data.

Contribution

It provides novel empirical evidence on the causality and interaction dynamics of VIX and its derivatives, addressing limitations of previous stochastic volatility models.

Findings

Bidirectional causality between VIX spot and implied volatility.

VIX spot tends to lag its own futures, indicating feedback effects.

Significant mean-reversion in the VIX and derivatives interactions.

Abstract

In February 2018, the VIX index has seen its largest ever increase and has lead to significant losses for some major volatility related products. Despite many efforts, the precise underlying reasons are yet to be discovered. We study the role of linear causality in the VIX index and its derivatives during January and February 2018. Due to the shortcomings of statistical inferences for stochastic volatility models, the dynamics of the volatility expectation index VIX remain controversial. Leveraging intraday data, we discover novel empirical results describing their interaction. We find bidirectional causality between the VIX spot and the implied volatility of Standard and Poors 500 options, suggesting a volatility feedback effect. The spot index tends to be lagging its own futures, while the vector autoregressions error correction mechanism reveals a significant mean-reverting equilibrium relationship. The evidence is consistent with recent theories indicating that implied volatility has stronger feedback than realized volatility. The paper reveals a retroactive information flow and highlights novel insights for the market microstructure of VIX derivatives and their related SnP 500 options.