Side Lesson 03: ETF Mechanics — Creation, Redemption, and the Tax-Efficiency Magic Trick

---

1. Why This Is Important

If you are following Horace's program at all faithfully, the single biggest position in your portfolio is an ETF — probably VTI, SPY, or QQQ. The reasoning is blunt: alpha is rare, the default is passive, and the cleanest way to own the US market is one of three ETFs that can be bought commission-free anywhere. So it is worth ten minutes of your life to understand the plumbing.

There are four reasons the plumbing matters:

**It explains why ETFs are tax-efficient and mutual funds are

not.* Two products that hold the same* portfolio at the same fee — VTI and VTSAX, both 100% Vanguard Total Stock Market — can produce wildly different after-tax returns in a taxable account because of one structural difference: in-kind versus cash redemptions. Once you see the mechanism you know which sleeve belongs in the brokerage account and which in the IRA.

It explains premium/discount. When a fixed-income ETF

prints on screen at -4% to its NAV during a panic, it is not "broken." It is doing exactly what its design says. Knowing that prevents the worst kind of forced sell.

**It explains why closed-end funds (CEFs) trade at persistent

discounts** while ETFs do not. Same wrapper category, very different mechanics. Recognising the distinction keeps you from buying a CEF at NAV and a year later discovering it is "supposed to" trade at -10%.

It explains the March 2020 bond-ETF dislocation — the

stress event bond ETFs passed even though the headline "ETF trading at a 5% discount" looked alarming. The headline was actually the ETF doing price discovery on a frozen underlying market. Understanding that puts you on the right side of the regulator's post-mortem.

This is one side lesson where the plumbing actually changes how you size positions. Worth the ten minutes.

---

2. What You Need to Know

2.1 Authorized Participants and the Creation/Redemption Cycle

An ETF is not a single thing. It is two markets stitched together.

The secondary market is what you see — Robinhood, Schwab, IBKR. You buy 100 shares of VTI from another investor through a public exchange. Cash for shares, exactly like a stock. The ETF issuer (Vanguard, BlackRock, State Street) is not involved in that trade.

The primary market is hidden from retail. A small group of large broker-dealers — Goldman, Morgan Stanley, Citadel Securities, Virtu, Jane Street, Flow Traders, Susquehanna — sign agreements with the ETF sponsor that make them **Authorized Participants (APs)**. There are usually 20-50 APs per ETF; in practice three or four do most of the volume on any given day.

The cycle works like this. The sponsor publishes a **creation basket** every morning before the open: the exact list of stocks (and weights) that one creation unit of the ETF holds. A creation unit is typically 25,000 to 100,000 ETF shares — call it $3-10 million of NAV.

When the ETF trades at a premium to NAV, an AP:

Buys the underlying basket of stocks on the open market.

Delivers that basket in kind to the ETF sponsor.

Receives a fresh creation unit of new ETF shares.

Sells those new ETF shares on the exchange at the premium.

The premium is the AP's profit. New supply hits the market, the premium compresses, and price re-aligns with NAV.

When the ETF trades at a discount, the cycle reverses:

The AP buys ETF shares cheaply on the exchange.

Delivers them back to the ETF sponsor.

Receives the underlying basket of stocks in kind.

Sells the basket on the open market at the higher fair value.

The discount is the AP's profit. ETF shares get retired, supply shrinks, price re-aligns with NAV.

This whole loop is automated, runs all day, and is invisible to you. The only direct evidence of it you ever see is the iNAV (intraday indicative value) that the exchange publishes every 15 seconds — a continuously-updated estimate of the basket value, against which the current bid/ask is checked. When iNAV and price diverge by more than a few basis points, an AP shows up.

2.2 Why ETFs Are Tax-Efficient (the In-Kind Trick)

Every fund that holds appreciated securities — mutual fund, ETF, hedge fund — has the same problem: when an investor wants out, the fund must somehow free up cash. If the fund sells appreciated stock to pay the redeemer, it realises a capital gain that is taxed at the fund level and the fund is required by law to distribute that gain to all remaining shareholders by year end. Loyal long-term holders pay tax on gains they did not ask the fund to realise. This is the core flaw of the open-end mutual fund structure.

ETFs sidestep this through the in-kind primary-market mechanism. When an AP redeems ETF shares for the underlying basket, the sponsor selects which specific tax lots of the underlying stocks to hand over — and naturally hands over the lots with the lowest cost basis (largest embedded gains). Those gains leave the fund permanently in a non-taxable in-kind transfer. Section 852(b)(6) of the Internal Revenue Code blesses this: an in-kind redemption is not a sale and triggers no fund-level capital gain.

The result is real, though for index mutual funds it is moderate rather than dramatic. Vanguard's VTSAX (mutual fund) and VTI (ETF) hold *the same portfolio at the same expense ratio* -- 0.04% on the institutional side. VTSAX has distributed long-term capital gains in years like 2014, 2015, 2018, 2020, and 2024 (each in the range of 0.1% to 0.7% of NAV); VTI has distributed essentially zero capital gains in its entire history. Over a 20-year hold in a taxable account at a 20% combined LTCG rate, the after-tax wealth difference is on the order of half a percent of terminal value -- a few thousand dollars on a 100k starting balance. For a typical active mutual fund pushing 5-10% of NAV in distributions during a turnover-heavy year, the same calculation produces tens of thousands of dollars of drag.

This is also why Vanguard's patented dual-share structure — VTI is literally a share class of VTSAX, sharing the same portfolio manager and the same trades — gave them an unusual edge for two decades. The patent expired in 2023. Other issuers are now launching ETF share classes of existing mutual funds, which is one of the more important structural shifts in the fund industry that nobody outside fund operations is paying attention to.

2.3 Premium/Discount, iNAV, and the March 2020 Bond ETF Dislocation

The arbitrage loop is fast for liquid US-equity ETFs. SPY, VTI, QQQ, IVV trade within 2 basis points of NAV essentially all day. You will never see a meaningful premium or discount on a megacap US-equity ETF in normal conditions.

Where it gets interesting is in less-liquid underlyings: high-yield bonds, emerging-market debt, municipal bonds, MLPs. When the underlying market is itself thin or temporarily frozen, the ETF price leads the underlying — the ETF trades continuously on the exchange, the underlying does not. The "discount" you see is often not a dislocation; it is the ETF price-discovering what the bonds are actually worth, ahead of the stale dealer marks that drive the published NAV.

That is exactly what happened in March 2020. On 12-13 March 2020, LQD (investment-grade bond ETF) printed at a 5% discount to its published NAV; HYG (high-yield) at 6%; MUB (munis) at 8%. The bond market itself had seized up: bid/ask spreads on individual investment-grade bonds blew out from a normal 5 cents to over a dollar, and many bonds simply did not trade. The ETFs did trade — and the price they printed turned out, in the post-mortem, to be a much better estimate of fair value than the official NAVs. When the Fed announced the corporate-bond facility on 23 March, the discounts closed and ETF prices led the recovery.

The SEC's 2020 Rule 6c-11 review concluded that ETFs functioned as designed and that the price-discovery role they played was useful during the dislocation. This is now the consensus regulatory view. Understanding it is useful because the next time a bond ETF prints at a -4% discount during a panic, you will recognise it for what it is — a feature, not a bug — and not panic-sell into the bid.

2.4 ETFs vs Closed-End Funds, and the New Fee Reality

The contrast with closed-end funds (CEFs) is sharp. A CEF issues a fixed number of shares once in an IPO and then those shares trade on an exchange like a stock — but with **no creation/redemption mechanism**. There is no AP. If the market price drifts away from NAV, no arbitrageur can pull it back. CEFs typically trade at -5% to -15% discounts to NAV persistently — often for years. Same wrapper category, completely different animal.

For most of the post-2000 era there was a real expense-ratio gap between ETFs and mutual funds — ETFs at 5-10 bps, comparable index mutual funds at 50-100 bps. That gap has largely closed for flagship index products. Today VTI and VTSAX are both 0.04%; IVV and VFIAX are 0.03% / 0.04%; BND and VBTLX are 0.03% / 0.05%. Where the gap remains huge is in active management: active mutual funds still average around 0.65% net of waivers; the median actively-managed ETF charges 0.35%. Active mutual funds still get fed retail-channel kickbacks (12b-1 fees, soft-dollar revenue-sharing) that the ETF wrapper structurally cannot pay.

The practical takeaway: for index exposure, the wrapper choice between ETF and mutual fund is now a tax-account-location question, not a fee question. VTI in the taxable brokerage account (in-kind tax shelter), VTSAX in the IRA (no tax to shelter, and the mutual fund auto-reinvests distributions to the penny).

---

3. Common Misconceptions

Misconception 1: "ETFs are tax-free." ETFs are not tax-free. Dividends are taxed (qualified or ordinary, same rules as stocks). Capital gains when you sell are taxed. What ETFs avoid is the fund-level capital-gain distribution that mutual funds dump on you in December. You control when the gain is realised, by choosing when to sell.

Misconception 2: "An ETF trading at a discount is broken." For liquid US-equity ETFs, even a 30 bp discount is unusual and means an AP is asleep. For less-liquid bond or emerging-market ETFs, a -2% to -5% discount during a stress event is normal and reflects the ETF leading the underlying in price discovery.

Misconception 3: "iNAV is the ETF's true price." iNAV is a model price computed every 15 seconds from the last prints of the underlying stocks. For a US-equity ETF holding liquid stocks during US trading hours, iNAV is reliable. For international-equity or bond ETFs whose underlying markets are closed, iNAV becomes stale and the traded ETF price is the better estimate of fair value.

Misconception 4: "Synthetic ETFs are the same as US ETFs." European synthetic ETFs use total-return swaps with a counterparty bank rather than holding the underlying physically. They have counterparty risk that physical ETFs do not. US-listed ETFs are physically backed by law (1940 Act). Stay in US-listed wrappers.

**Misconception 5: "Leveraged and inverse ETFs are just like regular ETFs but levered."** Leveraged ETFs use daily-rebalanced derivatives and suffer volatility decay — the multi-day return is not 2x or -1x the multi-day return of the underlying. SSO has under-performed frictionless 2x by several percentage points per year over long horizons (Week 37). Short-term tactical, not long-term hold.

**Misconception 6: "Vanguard mutual funds and ETFs are two different products."** For VTI/VTSAX and several other flagships, they are actually the same fund with two share classes — same portfolio, same manager, same trades. Until 2023 this was Vanguard-patented; now it is being copied across the industry.

**Misconception 7: "ETFs that track illiquid assets are doomed in a crisis."** The March 2020 stress test produced the opposite conclusion. Bond ETFs functioned as price-discovery mechanisms when the underlying markets froze. The ETF wrapper held up better than the underlying markets.

**Misconception 8: "Closed-end funds and ETFs are basically the same."** Same exchange-traded wrapper, completely different mechanism. ETFs have AP-driven creation/redemption that pins price to NAV. CEFs do not, and their prices drift to persistent -5% to -15% discounts. The acronym is similar; the product is not.

---

4. Q&A

**Q1: How do I check the premium/discount of an ETF before I trade?**

A: Every issuer publishes an end-of-day premium/discount on its fund page. For intraday checks, the exchange publishes iNAV under the symbol .IV (e.g. VTI.IV) on most data feeds. For US-equity ETFs you can usually skip this — premiums and discounts are below the bid/ask spread. For bond and international ETFs, checking before a large trade is worth thirty seconds.

Q2: Should I prefer market orders or limit orders on ETFs?

A: Limit orders, always, especially in the first 15 minutes after open and the last 15 minutes before close when iNAV can lag. For liquid US ETFs, place a limit at the displayed bid (selling) or ask (buying) and you will fill within seconds at NAV. Avoid market orders during fast tape — slippage on a thin mid-cap-sector ETF can be 10-20 bps.

**Q3: Why does VTI rarely distribute capital gains while VTSAX sometimes does?**

A: Same portfolio, but VTSAX has cash-redemption flow (an investor selling triggers the fund to sell stocks to raise cash, realising gains), while VTI has in-kind redemption flow (the AP takes the stocks, no sale, no gain realised inside the fund). When VTSAX sees outflows in a year of high embedded gains, the year-end distribution can be material. VTI almost never has this problem.

Q4: If VTI is more tax-efficient, why hold VTSAX at all?

A: Two reasons. VTSAX auto-reinvests dividends and distributions to the penny with no fractional-share friction — useful in a 401(k) or IRA where tax efficiency does not matter and you want the cleanest accounting. VTSAX also accepts dollar amounts ("invest 500"), VTI accepts share amounts ("buy 4 shares"); for automated regular contributions the dollar-amount interface of the mutual fund is simpler. The tax rule still says: in a taxable brokerage account, default to the ETF.

Q5: What are creation units and why are they so large?

A: A creation unit is the minimum block size at which an AP can interact with the ETF sponsor — typically 25,000 to 100,000 ETF shares (3-10 million of NAV). The size keeps operational and custody costs of physical creation/redemption manageable. APs break those blocks down into the much smaller lots retail investors trade.

Q6: Can the ETF mechanism break? What would that look like?

A: It can break in two ways. (1) APs go on strike — refuse to arb because they think they cannot exit the underlying. This is what regulators worry about for thin underlyings. (2) The underlying market itself ceases to function — March 2020 was a near miss. In both cases the ETF is doing what it should: showing the world that the underlying is priced incorrectly. The fix is to fix the underlying market, not to ban the ETF.

Q7: What is a "synthetic" ETF and should I care?

A: A synthetic ETF (mostly European) holds a swap contract with a bank counterparty that promises to pay the index return, rather than holding the underlying stocks. It introduces counterparty risk. The default is US-listed ETFs only — that constraint automatically rules out synthetic ETFs.

Q8: Why do some ETFs charge 0.03% and others 0.95%?

A: Index ETFs that track a broad, well-licensed benchmark have license fees of essentially zero and operate at scale, so 0.03% is achievable. Active ETFs, thematic ETFs, single-country emerging-market ETFs, leveraged ETFs, and currency-hedged ETFs all carry higher fees because of licensing, hedging, or active management costs. The rule still holds: alpha is rare, and the cheap ones almost always win after 10 years.

Q9: What is the "ETF tax-loss harvesting partner" trick?

A: To realise a tax loss while staying invested, you sell one fund and buy a "substantially-similar-but-not-identical" fund. ITOT and VTI both track total-US-stock-market indices but are different funds (one BlackRock, one Vanguard, slightly different indices), so swapping them is not a wash sale. Side Lesson 4 covers this in detail.

**Q10: What is the most important thing to remember from this lesson?**

A: Two things. In a taxable account, prefer the ETF wrapper to the equivalent mutual fund — the in-kind redemption mechanism gives you a structural after-tax edge that compounds over decades. And when an ETF prints at a discount during stress, ask whether the underlying market is itself functioning before you conclude the ETF is broken — almost always the ETF is right and the NAV is stale.