ETFs In Chaos: Shock Absorbers

There’s a lot of misinformation about how ETFs work in times of volatility. We just had a master class.

Reviewed by: Dave Nadig
Edited by: Dave Nadig

The No. 1 question I got asked in the month of October was, “How’d ETFs do?” What’s behind that question was a decade of hand-wringing by pundits claiming ETFs would cause, or exacerbate, the next big crisis.

Instead, what I think we’ve seen is ETFs acting as shock absorbers for the market—putting bids underneath falling securities. If nothing else, we’ve seen firsthand some great examples of precisely how creation and redemption work to keep ETF pricing in line in times of stress.

Let’s start first with the world’s most liquid security—the SPDR S&P 500 ETF Trust (SPY)Most ETF investors know that for all ETFs, even SPY, there are two “values” at any given moment in time. There’s the price at which it’s trading, and there’s a “fair value”—the actual net asset value based on all the underlying securities; in the case of SPY, the value of all the stocks in the S&P 500.

These two values don’t always match perfectly, so investors can monitor that “fair value” through the intraday NAV, or iNAV. iNAV has lots of problems, but in the case of big, liquid funds and big, liquid securities, it works pretty well.

A Tale Of 2 Days

Here’s the difference between SPY’s trading price, and the iNAV, during the most volatile parts of the last month:



A few things to note: This is actually pretty typical. SPY trades currently in the $270 per share range, and it can be “mispriced” by 10-30 cents regularly—that’s about 0.07% plus or minus few hundredths of a percent if you’re doing the math, and that falls among the smallest perceived premia/discounts in the market.

I say “perceived” because there are a lot of factors that go into this: differences in the timing of when stocks trade versus when the ETF trades. The spreads on the underlying stocks themselves are often much wider than that 0.07%, so there’s a lot of “wiggle room” in this analysis.

But what we do know is when it makes sense for an authorized participant to do something about it. Anytime the stocks in the S&P 500 seem “cheap” relative to the price of SPY, they will buy the stocks, and sell the ETF. If the stocks seem “expensive” relative to the ETF, they’ll sell the stocks, and buy the ETF.

If, at the end of the day, they end up with a lot of stock, and have sold a lot of SPY they don’t actually own, they do a creation, trading the stock for the shares of SPY, and they go home happy.

So in this recent volatile window, what went on with SPY in terms of creation/redemption:



Clearly, a little bit of each. A few huge caveats: Flows data is self-reported, and not always reported the same way by all issuers. We know from experience that the flows in SPY as we report them here (from FactSet data) are in fact a day off. That is, the inflows you see on Oct. 11 on this chart, of roughly $2.5 billion, represent orders placed for creation on Oct. 10.

If we go back to Oct. 10 on the first chart, what do we see? Well, SPY looks like it traded above fair value for most of the day. So, you’d expect that APs would sell the ETF and buy the underlying stocks, making new shares at the end of the day—positive inflows. And that’s what happened.

If you move ahead a day to the big outflows on Oct. 12, you can see that’s a response to the discounts from Oct. 11. In other words, SPY was trading under fair value, so the AP bought shares of SPY, and sold shares of stock and did a redemption at the end of the day.

Here’s a big, important point: Both of these were bad days. These were days when the VIX went above 20. The role of SPY here—which is used as a trading vehicle extensively—was to actually absorb demand, in whatever way the market wanted to express it. By putting a bid below the stocks, or an offer over them, I’d argue that SPY’s presence in the market was acting as a volatility reducer, not an accelerator

So let’s look to a different S&P 500 fund, one that we don’t think of as a trading vehicle, but as a buy-and-hold investment: the Vanguard S&P 500 ETF (VOO). While VOO certainly trades well, it’s more notable for being a bit cheaper than SPY (0.04% versus 0.09%) and crushing it in flows lately ($15 billion in new money for VOO, $22.5 billion in outflows from SPY year-to-date).

Here’s the premium/discount over the same window:



While VOO trades substantially less than SPY, it generally experiences about the same swings. VOO has a slightly lower price, and trades in a slightly tighter band to fair value. What I see on the margin above is quite a few days throughout October where the stocks were oversold versus the price of VOO itself. It vacillates, for sure. And as such, the creation and redemption activity goes back and forth as well.



Here again, I think we’re seeing the “shock absorber” effect. If the stocks are “oversold,” then APs are going into the market to buy them while selling the ETF; if they’re “overbought,” they do the opposite. The net effect is, again, putting bids under the underlying stocks on the worst days.

Backward Bonds

Is this a perfect analogy for all ETFs in all asset classes? Of course not.

The dynamics of the underlying markets make for a very different story in bonds, for instance, where the ETFs have become the price discovery mechanism. iNAV for a bond ETF, like the iShares iBoxx USD High Yield Corporate Bond ETF (HYG), is inherently fictitious. It’s based on bond pricing services that try and create real-time indicative prices for junk bonds that may not have traded in hours, or even days.

Those pricing services themselves use the price of ETFs as one of their inputs. As such, any perceived premium in the price of a junk bond ETF is in fact just a recognition that the market has moved outside the ability of the pricing services to catch up.

Here, for example, is how HYG looked in this same volatile window:



Here the story seems even clearer when you look at the flows:



The sustained four days of premia in the ETF here aren’t the result of the “bonds being cheap” so much as they’re the result of the ETF trading furiously enough, for long enough, to establish that the APs could in fact do creations and make money doing so to meet demand. These weren’t big move days for HYG—it moved less than a percent each day. But volumes for all bond ETFs (HYG included) went through the roof in October.

Moral Of The Story

Look, I’m the first one to acknowledge this is anecdotal analysis of three ETFs, not an academic study. I’m also the first to point out that iNAV, ETF fund flows and premium/discount analysis have loads of problems—I’ve written about them all. But we work with the data the market gives us on any given day, and October is a fishbowl in which we can look to see how the ecosystem is working.

And I take comfort that in the dozens of ETFs I’ve checked on in October, everything seems to be working just like it’s supposed to: APs are stepping in, making markets, arbitraging out the price discrepancies, and yes, providing a kind of “liquidity buffer” for big market movements.

Dave Nadig is managing director of He can be reached at [email protected].

Prior to becoming chief investment officer and director of research at ETF Trends, Dave Nadig was managing director of Previously, he was director of ETFs at FactSet Research Systems. Before that, as managing director at BGI, Nadig helped design some of the first ETFs. As co-founder of Cerulli Associates, he conducted some of the earliest research on fee-only financial advisors and the rise of indexing.