Exchange-traded funds (ETF) have soared in popularity in recent years, and it's impacted the stock market
You know all those predictions that we're in a "stock picker's market"? Well, guess what: We're not. In fact, by one definition, we're in less of a stock picker's market than at any time ever.
Bespoke Investment Group tracks something called "All or Nothing Days." The definition is that it's a day where 400 of the roughly 500 names in the S&P 500 Index (SPX) move in the same direction. As per their numbers, there were 159 trading days from the beginning of 2015 through Aug. 19, and in that time frame, we saw 13 "All or Nothing Days." And then, in the subsequent 15 trading days alone, we saw 11 more. To which they say:
"While it is common to see an uptick in the number of all or nothing days in the S&P 500 as volatility picks up, the frequency of all or nothing days in the S&P 500 over the last 15 trading days is unheard of. Using our breadth data going back to 1990, we have never seen a 15-trading day period where the S&P 500 saw as many or more all or nothing days than it has in the current period. The prior record was back during the financial crisis when we saw ten over a 15-trading period in late 2008."
Well, this all-or-nothing number is a good proxy for correlation. Correlation basically equals volatility, at least as far as an index is concerned. We can simplify the realized volatility of an index into two parts:
- The volatility of the components themselves, and
- The degree to which the movement in the components correlates.
If the components are going nuts, but they're moving in 500 different directions, index volatility will be on the low end. Conversely, if they're all modestly volatile, but all moving in the same direction, the index will have some volatility. So it's safe to say that the volatility we've seen is a function of increased correlation. And that makes intuitive sense. The easiest way to adjust risk while the market is in panic mode is to sell futures or index exchange-traded funds (ETF). You can always tweak the specifics later. Which brings us to their further point:
"As shown, all or nothing days were few and far between from 1990 through 2000. The reason for this is that the uptick in frequency of all or nothing days has coincided pretty closely with the increased popularity of ETFs, as buying or selling in securities like SPY causes buying or selling pressure in each of the securities compromising the S&P 500. While ETFs have been beneficial for investors looking for a low cost way of gaining exposure to the broad market, one of the unintended consequences has been that the daily correlation of individual stocks has increased."
And that leads to an odd conundrum. The growth over time of ETFs has absolutely altered stock behavior in and of itself. It stands to reason that correlation would increase as ETFs, as a share of the marketplace, has increased.
Yet, volatility has not increased over time. It ebbs and flows, and ultimately mean-reverts now at about the same levels it did 10 years ago and 20 years ago. Here's a chart of the median for the CBOE Volatility Index (VIX) by year, going back to 1990:
Same as it ever was. So maybe these correlation spikes beget faster vol spikes in the moment. In fact, they most certainly do. But over time, the growth of ETFs doesn't seem to have much impact on vol. I'm not entirely sure, though my best guess is that since it's not the "only" thing that impacts vol, it's offset by something that serves to decrease vol over time. Automation? Penny and micro-penny spreads? Perhaps. It's the subject for another day.
Disclaimer: Mr. Warner's opinions expressed above do not necessarily represent the views of Schaeffer's Investment Research