How Bond ETFs Work

Buy a share of a stock ETF and everything inside it trades on the same exchange you just used. Buy a share of a bond ETF and almost nothing inside it does.

The bond market has no central exchange. It runs over the counter, dealer to dealer, in sizes built for institutions, and on any given day most individual bonds do not trade at all. A bond ETF holding a thousand of those bonds trades every second the market is open, at a spread measured in pennies.

That gap between the wrapper and its contents is the whole subject. I run an ETF business, so here is how it actually works.

What is inside one

A bond ETF holds bonds. Treasuries, corporates, mortgages, municipals, or some combination, depending on what it says on the label.

The complication is that "a bond" is not one thing the way "a share of Apple" is one thing. Apple has one common stock trading in one place. A large corporate issuer can have twenty or thirty separate bonds outstanding, each with its own coupon, its own maturity, its own covenant package, and its own price. The universe is enormous and most of it is quiet on any given day.

Steve Laipply, Global Co-Head of iShares Fixed Income ETFs at BlackRock, came into the category from the dealer side and described the market he grew up in on Behind the Ticker. Individual bonds trade over the counter, he said, and "before you had trace, it was even" harder to understand price discovery. TRACE, the reporting system that arrived in the early 2000s, improved that. It did not turn the bond market into an exchange, and the ETF still has to be built on top of a market that quotes by appointment.

The odd lot problem

Here is the first thing a bond ETF solves, and it is the one advisers feel directly.

Bond pricing is sensitive to size in a way equity pricing is not. Institutional-sized trades get institutional pricing. Small ones get whatever a dealer feels like quoting on a position they do not want.

Eric Lutton, Chief Investment Officer at Sound Income Strategies, explained why his firm launched a fixed income ETF at all, and the reason was small accounts. With bonds, he said, you do not want to "have a one or two lot in your portfolio", because if the client ever left and you were trying to sell it, they would get very poor prices.

A fifty thousand dollar account cannot assemble a diversified bond portfolio out of individual bonds without paying for the privilege on the way in and again on the way out. The fund transacts in institutional size and the investor buys a share of the result. That is the plainest case for the wrapper, and it has nothing to do with what anyone thinks about interest rates.

Somebody had to build the plumbing

None of this arrived for free.

Joanna Gallegos co-founded BondBloxx after two decades in the category, starting at Barclays Global Investors in the early days of iShares. She was on the team that launched the first high yield ETF, and her account of it is the best short description of what the wrapper actually required:

"When we launched HYG in 2007, there just wasn’t the bond market infrastructure to match sort of the equity infrastructure that ETFs operated in. And so there’s a lot of work that had to be done to connect all of those points. Literally going to different desks at different big firms and explaining them how to price a bond portfolio of so many bonds and then how to get that working."

That is the piece most explanations leave out. An equity ETF inherited a market that already produced continuous prices. A bond ETF had to be wired into one that did not, desk by desk.

The inversion

Now the part that surprises people, including people who work in bonds.

The natural assumption is that a fund is downstream of its holdings. The bonds have a price, the fund adds them up, the fund’s price follows. In fixed income, under stress, it frequently runs the other way.

Laipply has published research on this for years and gave the cleanest version of it on the show, using the largest investment grade corporate bond ETF during the spring of 2020:

"LQD on certain days traded, you know, 90 100,000 times on exchange, the top holdings in LQD, during that same period of time may have traded maybe like a dozen times. So which price are you going to believe something based on, you know, 90,000 times or 100,000 times or something that’s, you know, a dozen times, right."

Read that again. The fund printed a price roughly a hundred thousand times in a session. Its largest underlying positions printed about twelve.

A fund’s net asset value is struck once a day off pricing services. Laipply is careful to say that everyone works hard to make NAV as accurate as possible, and then adds the thing that matters: "a lot of days, things aren’t trading", and the pricing services are forced to estimate.

The exchange price is a completed transaction between two parties with money at risk. So in a genuine dislocation the market price leads, the gap to net asset value widens, and the underlying market catches up afterward. In his words, "the market price does lead", especially in stress markets, and eventually the two converge.

An investor who sees a bond ETF trading below its stated NAV and reads that as the fund being broken has the causality backwards. Often the fund is the only thing in the room quoting a live price.

A thousand bonds, a penny wide

The second consequence is execution.

I put the most common objection to Laipply directly, the one advisers raised constantly through 2022: why do I need a bond ETF when I can buy individual bonds and hold them to maturity and never worry about the mark? His answer was about what you can physically transact:

"An LQD, for example, right, you have well over 1000 bonds, a penny wide, you’re not going to be able to do that trade in the underlying market full stop ever, it’s really hard to do."

Assembling a thousand-bond portfolio directly means a thousand separate negotiations at retail size. The fund compresses that into one order on an exchange. And the institutional workaround that grew up alongside it runs on the same rails: even now on the institutional side there are things called "portfolio trades", and those trades are based on an ETF powering them.

Three honest limits

A broad bond fund has no maturity date. An individual bond matures and pays you par, which is what lets people say they will hold through a drawdown and be made whole. A broad bond ETF is a perpetual ladder: bonds mature inside it, the proceeds buy new ones, duration stays roughly constant, and there is no date on which you get par back. Some funds solve this deliberately with defined-maturity structures. Most do not, and it matters most to somebody funding a known liability on a known date.

Duration is the exposure that shows up in the mark. The traditional core index is investment grade across treasuries, agencies, agency mortgages, corporates and securitized paper, and Laipply puts its duration "pushing sort of close to six". Rates move that number against you or for you long before credit does anything at all.

Broad bond indexes are sampled, not replicated. A broad index can contain many multiples of what any fund holds. The fund owns a representative subset chosen to match the index on duration, sector and credit. The method works well, and it is worth knowing it is the method.

Where the return comes from

One more piece, because it changes how a bond fund’s chart should be read.

Gallegos, on high yield, went back to what she called bond math: the coupon income has been the primary driver of long-term returns, and a large coupon absorbs a lot of price volatility along the way.

Price moves get the attention. The coupon does the work, and it accrues every day regardless of what the price did. A fund whose price is flat across a year has not necessarily done nothing.

Four things to check

What it actually holds. Bond ETFs publish holdings daily like every other ETF. Treasury, corporate, high yield, securitized and municipal funds behave nothing alike, and several of them share adjectives in their names.

Duration. The best single predictor of how the fund moves when rates move. It is published. Compare it to the horizon you are actually investing over.

Credit quality. Investment grade and high yield are different risks with different relationships to equities. Gallegos built an entire product line on that distinction, splitting high yield by rating and by sector, because "trading through the credit spectrum" is how institutional investors already think about where their risk sits.

The spread under stress. A bond ETF’s spread widens when the underlying market thins out. That is not a defect in the fund. It is the underlying market showing through the wrapper. Size the expectation for the day you need to sell, not the day you buy.

The part worth keeping

A bond ETF is a way to hold a market that was never built for individual investors to hold. It transacts in institutional size, prices on an exchange all day, and in a dislocation it tends to say where the market is before the market gets around to saying so.

The wrapper is well tested and it does what it claims. The questions worth asking are about what somebody put inside it, how long its duration is, and what it is being asked to do in the portfolio.

This is educational content and not investment advice. It is not a recommendation regarding any security. Fund names appear only as examples of the mechanics described. Investing involves risk, including possible loss of principal. Past performance does not guarantee future results.