Premiums and Discounts

Definitions, and one measurement trap

Premium/discount = (market price − NAV) / NAV. Issuers publish it daily; fund pages chart its history. The trap in the measurement: the two numbers carry timestamps. NAV is struck once, at the underlying market's close, using those closing prices; the ETF's own closing price is set wherever it trades, possibly hours later, incorporating everything that happened since. Compare a 17:30 CET Europe-listed price of a US-equity ETF with a NAV built from yesterday's 22:00 CET New York close, and you will "measure" a premium that is nothing but six hours of news. A large share of the scary premium/discount charts on the internet are this artifact: not mispricing, mis-clocking.

The three honest causes

1. Asynchronous markets: the forecast vs the photograph. A Japan-equity ETF trading in London at noon prices Japanese companies while Tokyo sleeps. Its market price embeds the MM's best fair-value estimate (futures, FX, correlated markets, Chapter 32's craft); its official NAV still shows Tokyo's last close. The resulting "premium" of ±0.5–1% on volatile days is the ETF leading its own NAV, and studies of the next Tokyo open confirm it: the underlying market opens toward where the ETF already traded. Verdict: not a cost, a time machine. Chapter 12's awake-underlying rule remains the practical response.

2. Basket friction: the conversion toll. For funds whose baskets are expensive to trade (corporate bonds, EM local markets), arbitrage only fires when the gap exceeds the cost of the round trip. The premium/discount therefore breathes inside a no-arbitrage band roughly the width of the basket's true trading cost: a corporate-bond ETF drifting between +0.3% and −0.3% is showing you, in real time, what entering or exiting its market actually costs, information the fund's TER never contained. Systematic feature: such funds also sit at small persistent premiums in calm times, because creations (which require buying bonds at offer prices) cost more than the NAV's mid-price marks: again, honest accounting, not error.

3. Stress: price discovery in public. March 2020, the canonical case: major bond ETFs traded at discounts of 3–6% to NAV while the underlying bonds barely traded at all. The naive reading was "ETFs broke." The examined reading, now the mainstream one among regulators and academics: NAVs were built from stale, unexecutable quotes, and the ETF price, set by thousands of actual transactions per minute, was the real price of bonds that week. The discount was the messenger. Evidence: when the Fed later bought actual bonds, it paid prices near where the ETFs had been trading, and NAVs converged down to the ETFs as pricing services caught up. The wrapper functioned as the bond market's only real-time tape. Chapter 82 gives the episode its full case study; the portable lesson lives here: in a crisis, distrust the NAV before you distrust the traded price.

Reading the instrument panel, practically

• Check the history, not the snapshot. Issuer pages chart 1–5 years of premium/discount. A broad equity fund hugging zero, a bond fund breathing in a ±0.3% band, an Asia fund sawtoothing with time zones: all healthy signatures. The unhealthy one: a fund whose gap trends or sits persistently wide versus peers on the same basket, which flags weak arbitrage coverage (Chapter 32) or a structural issue worth understanding before owning.
• Never pay a stress premium or sell a stress discount blind. If a fund is at +2% because its market is closed or seized, a market order gifts the gap to the arbitrageur. The Chapter 15 rules compound here: limits, patience, and for large moves, waiting for the underlying to reopen costs hours and saves percent.
• Use it for timing execution, never allocation. The gap tells you about the plumbing's current toll, nothing about whether the asset is cheap. "Buying the discount" on a time-zone artifact buys you nothing; on a stress discount it buys you the crisis you presumably wanted or feared anyway, at fair-that-day prices.

The comparison that settles the concept

Chapter 4's investment trust trades at −8% for a decade because no mechanism exists to close it. An ETF's discount, even 2020's dramatic one, closed in days, because Chapter 31's window makes persistent gaps into standing paychecks. Same word, "discount"; different physics. The premium/discount chart is, in the end, a live audit of whether the physics is working, and for three decades, across every stress test, its verdict has read: bent, priced, closed.

A worked example

Three screenshots, one Tuesday:

• 09:05 CET: your Europe-listed S&P 500 ETF shows +0.6% premium to NAV. Diagnosis: NAV is last night's New York close; US futures are up 0.6% overnight. Action: none; that's a clock, not a cost.
• 14:30 CET: an EM local-currency bond ETF you're buying shows +0.35%, its five-year chart breathing ±0.4%. Diagnosis: the conversion toll of its basket. Action: buy with a limit as planned; the toll is the asset class's, and every vehicle for it pays one.
• A crisis Friday: a high-yield ETF prints −4% while its NAV eases −0.5%. Diagnosis: stale NAV, real panic. Action per your written policy: if rebalancing says buy, the discount means you're buying at the actual market price, likely better than any fund transacting at tomorrow's marked-down NAV; if nothing in the policy triggers, close the app: the gap is information, not instruction.