With VIX ending Friday at its lowest weekly close ever, lowest monthly close ever, and lowest quarterly close ever - after the quietest September stock market in history - SocGen warned last Friday that the current situation is a "dangerous volatility regime" citing the strong mean-reverting tendency of uncertainty as a big reason for investors to brace themselves for trouble ahead.
Of course, judging by the new record short in VIX futures - extending last week's surge - the speculative public is as levered-long and complacent as it has ever been...
What could go wrong? SocGen's Arthur van Slooten explained late last week: "compare that with dancing on the rim of a volcano. If there is a sudden eruption (of volatility) you get badly burned"
Indeed, if CNN's Fear and Greed Index is anything to go by, investors are close to the 'most extreme' levels of greed in history...
However, it's not just volatility that is collapsing. Correlation has crashed, as JPMorgan's Marko Kolanovic recently pointed out.
Decline in correlations and parallels to 1994 and 2001
Over the past year, correlation of stocks and sectors declined at an unprecedented speed and magnitude (see figure below). A similar decorrelation occurred on only two other occasions over the last 30 years: in 1993 and 2000. Both of those episodes led to subsequent market weakness and an increase in volatility (in 1994, and 2001). The current decline in market correlations started following the US elections and was largely driven by macro (rather than stock-specific) forces. Expectation of fiscal measures, deregulation and higher interest rates set in motion large equity sector and style rotations. For instance, the correlation between Financials and Technology dropped to all-time lows (similar level during the tech bubble). The correlation between equity styles also dropped (e.g., Value was lifted by rates, and Low Volatility was impacted negatively). Declining correlations pushed market volatility lower (see here), and the ~25% market rally further suppressed correlations and volatility. To investigate what are potential implications for the future price action, we look at the 1993 and 2000 decorrelation events.
- 1993/1994: Following the 1990-91 recession, interest rates declined and the market rallied. By late 1993, the market reached its highs (60% above recession lows) and volatility plummeted (VIX hit a record low on 12/22/1993). This also marked the low point of equity correlations. As interest rates increased in 1994, the market experienced a ~10% correction and posted a negative return for the year. Volatility and correlation increased, but the crisis was contained given the acceleration of growth (US GDP increased from 2.6% to 4.3% in the first half of 1994), and subsequent decline in bond yields.
- 2000/2001: Following the 1998 crisis (LTCM, Russia), the market recovered and continued to rally. When the internet bubble was inflated, the market was 60% above 1998 lows. This period was marked with a strong decoupling of sectors (e.g., tech vs. financials), distorted valuations, elevated volatility and gradually rising interest rates. It ended with the tech bubble in March 2001, which marked the low point of equity correlation and start of recession. Subsequently, the market declined ~30%, bottoming in late 2002.
The current episode of correlation decline shares some similar features with both 1993 and 2000. The decline of correlation was in part driven by the market rally and elevated valuations; after a period of falling, interest rates are expected to rise (as in 1993), sector valuations (e.g., Internet) and sector rotations play an outsized role in market price action (similar to 2000), and record low levels of volatility increased the level of risk taking (as in 1993). Normalization of monetary policy will most likely lead to an increase of correlations and volatility, and that will at some point result in market weakness. While it seems that the 1993/1994 analogy is more appropriate (implying an orderly price action), investors should be aware of hidden leverage and tail risk of a more significant correction, such as the one in 2001.
Kolanovic concludes that "strategies that boost leverage when volatility declines, such as option hedging, CTAs and risk-parity, share similar features with the dynamic ‘portfolio insurance’ of 1987,” which “creates a ‘stop-loss order’ that gets larger in size and closer to the current market price as volatility gets lower.” Additionally, growth in short-vol strategies suppresses both implied and realized volatility, and with volatility at all-time lows “we may be very close to the turning point.”