What Signals a Coming Market/VIX Blow Up?

Derek Horstmeyer, Hugh Holtman, Ajeet BondugulaAdvisor Perspectives welcomes guest contributions. The views presented here do not necessarily represent those of Advisor Perspectives.

Investors and asset managers are often looking to markets to see if there are any early warning signs before a market blows up. Some investors swear by metrics such as credit expansion, IPO mania, or even meme stocks as early evidence that something is about to go wrong. Correlations between assets are often cited as an overlooked warning sign.

Because of this, we decided to examine six different major volatility events over the past 30 years and see what happened to asset correlations before and during these events. To examine a full spectrum of changing correlations, we could hopefully find early indicators before market/VIX blowups in the U.S.

What we found in general was that excessively high correlations between equity classes are typically present leading up to a VIX event, and excessively negative correlations between gold and equity also is also a leading indicator. Diversification benefits in general unravel a bit during the actual market blowup, with most assets becoming more positively correlated.

The Data Inputs

To study this, my research assistants (Hugh Holtman and Ajeet Bondugula) and I pulled data for all asset prices going back 30-plus years. Included in this full asset pricing data set were commodities, debt, and equities. For commodities, we included gold, oil, natural gas, silver, platinum, corn, wheat, and coffee. For debt, we included high-yield debt, U.S. corporate debt, intermediate U.S. Treasuries, U.S. long-term Treasuries, the MSCI World Bond Index, TIPs, munis, GNMA, and short-term Treasuries. For equities, we included small-cap stocks, large-cap stocks, international equity, and U.S. growth equities.

We identified six major VIX spikes as our volatility events. In these VIX spikes, we looked at events during which the VIX doubled in less than three months. The final study included VIX events in 1998, 2002, 2008, 2010, 2011, and 2020.

Next, with this data in hand, we put together correlation matrices for the assets three months prior to the VIX event and an equivalent correlation matrix for the assets during the month of the VIX spike. We then examined how correlations changed in this three-month period for all VIX spike events.

Below presents a conglomerate picture of all VIX events combined into one correlation matrix. This conglomerate picture details the change in correlations from three months prior to the event date (month of the VIX event).

Source: Authors and Morningstar Direct