Why does an ETF have two liquidities, not one?
The volume lesson in foundations taught you to read a stock's liquidity from its volume and its spread. An ETF has that layer, and a second one underneath that most holders never look at.
Layer one: the fund's own market
The units trade on an exchange with a bid, an ask and a daily volume, like a share. A large fund on liquid holdings has a spread of a cent or two and hundreds of millions of dollars of daily turnover. A small fund on the same holdings may trade a few thousand units a day with a spread of several tenths of a percent. That is the liquidity you pay for directly, every time you trade.
Layer two: the holdings
Behind the units sits the basket, and the creation mechanism from unit 1 means the fund's units can be manufactured from the basket on demand. So a thinly traded ETF on very liquid shares is more liquid than its own volume suggests: a large order can be filled by a participant creating new units, at a price close to the cost of buying the basket. The fund's spread is bounded by the basket's spread plus the participant's margin.
The reverse is the important case. An actively traded ETF on illiquid holdings is exactly as liquid as its holdings, whatever its own volume says. When the holdings stop trading, the participants cannot price the basket, the arbitrage stops, and the units trade at whatever the crowd will pay — the March 2020 discount from unit 1, seen from the liquidity side.
The rule for reading it
Ask two questions of any fund before a large order. How does the fund itself trade — spread and volume, readable from the quote. How do its holdings trade — and if the answer is "over the counter, by appointment", the first answer was decorative.
For a fund of S&P 500 shares, both answers are good and the second one makes the first one better. For a fund of high-yield bonds, the first answer looks fine on an ordinary day and the second is where the risk lives. The market-makers lesson called liquidity fair-weather; a bond ETF is where that becomes measurable.
What the wrapper adds
It is fashionable to say ETFs hide the illiquidity of what they hold. The opposite is closer to true. A bond ETF gives a continuous, public price for a basket of bonds that would otherwise trade by telephone, and in March 2020 that price was the most honest number available: it fell before the NAV did, because the NAV was built from stale quotes. The wrapper did not create the illiquidity; it made it visible, at a discount that was the cost of visibility. The mistake is treating the unit's own volume as a promise about the holdings.
In the data
Layer one is readable from the fund's own quote. A large equity fund and a large high-yield bond fund, side by side:
Bid, ask and last trade: everything the first question needs. Layer two is the same quote for what the funds hold, and for the bond fund there is none to show. Its bonds trade over the counter, dealer to dealer, with no public bid and ask of this kind. That gap is the lesson: if you cannot see the holdings' liquidity, you are trusting the fund's.
Try it now
From the quote above, compute each fund's spread (ask minus bid) as a percentage of its price. On a calm day the two look similar.
Now the bond fund across its life, with the measuring tool:
Find the sharpest two-week fall and measure it. Then find the same fortnight on an S&P 500 fund:
- Both fell. Say which of the two had a second liquidity layer that stopped working, and how you would have known it from the quote alone — or whether you could have.