Premium, Discount, and Tracking Error: Three Gaps, Three Causes

The distinction is straightforward once stated. Tracking difference is the gap between the fund's return and the index's return — a portfolio-management outcome, driven by costs and implementation. Premium or discount is the gap between the ETF's market price and its net asset value — a trading outcome, driven by supply, demand, and the limits of the arbitrage mechanism. A fund can track its index beautifully while trading at a discount, or track poorly while pricing perfectly. This article takes each apart, then explains why neither figure is a screening criterion that identifies good funds — which is how they are most often misused.

Tracking difference and tracking error: two words, two meanings

The terminology is genuinely confusing and worth fixing first, because the two terms are routinely used interchangeably and mean different things. Tracking difference (sometimes tracking error in casual usage) is the simple return gap over a period: index returned 8.40%, fund returned 8.14%, difference −26 basis points. It has a sign, and the sign matters — a fund can, occasionally, beat its index, usually through securities lending revenue or favourable tax treatment. Tracking error in its technical sense is the volatility of that difference — the standard deviation of return gaps over time, which measures consistency rather than magnitude. A fund that lags by exactly 20 basis points every year has a large tracking difference and near-zero tracking error; a fund that alternates between +50 and −50 has zero average difference and high tracking error. Both figures are informative and they answer different questions: how much does it cost me, and how predictable is it? Now the causes, which are almost entirely unglamorous. The ongoing charge is the largest and most predictable component — it comes out of the fund and not out of the index, so it lags by roughly its own size before anything else happens. Transaction costs from rebalancing arise whenever the index changes and the fund must trade to match. Cash drag — dividends received between reinvestment points, and inflows awaiting deployment — sits idle while the index assumes full investment. Sampling differences matter for funds that do not hold every constituent, as the index-fund article described. Withholding tax on dividends is often unrecoverable in part, while many indices are computed on a gross-dividend basis the fund cannot achieve. Securities lending revenue pushes the other way, adding return and sometimes offsetting charges entirely — the subject of its own article. And currency and timing effects arise where the fund and index are valued at different moments or in different currencies. One important qualification: comparing a fund to the wrong version of its index produces a fake tracking difference. Indices are published in price, net total return, and gross total return variants, and a fund tracking the net-return version will look dreadful against the gross-return version for reasons that have nothing to do with the fund. Always check which index version the fund states as its benchmark.

Premium and discount: a trading gap, not a management outcome

Now the other gap. An ETF's market price is set by buyers and sellers on an exchange; its NAV is the value of what it holds. When price exceeds NAV the ETF trades at a premium; below, at a discount. The creation-redemption mechanism exists to keep this gap small, and for ETFs holding liquid, continuously traded securities it typically is — fractions of a percent, fleeting. Four situations produce larger or more persistent gaps, and all four are about the arbitrage being harder rather than about the fund being worse. Illiquid underlying holdings. If assembling the basket is expensive and uncertain — thinly traded bonds, small-company shares — the arbitrage is only worth doing at a wider gap, so the gap persists. Stale NAV from closed markets. An ETF listed in Europe holding Asian shares trades all afternoon while its underlying market is shut: the NAV reflects Asian closing prices, while the ETF price reflects the market's live view of what those shares will be worth when Asia reopens. A "premium" here may be the ETF price being more current than the NAV, not a mispricing at all — which is why premium and discount figures for ETFs over closed markets need careful reading rather than alarm. Stress conditions. When underlying markets are disorderly, both NAV estimation and arbitrage become harder, and gaps have widened materially in documented episodes — most notably in fixed-income ETFs during acute market dislocation, where the ETF price arguably provided a more honest live signal of bond values than the model-derived NAVs did, an interpretation that generated substantial debate. And structural constraints: capacity limits on commodity or currency exposures, or funds where new creations are suspended, can sustain premiums indefinitely, since the arbitrage that would close them is unavailable. Practical guidance: for the reader, the premium or discount at the moment of dealing is a real cost or benefit, and a persistent pattern of wide gaps is worth understanding before dealing in size. Using limit orders rather than market orders, and avoiding the opening and closing minutes when spreads and gaps are typically widest, are mechanical observations about execution rather than trading advice.

Why neither figure is a fund-quality score

Both numbers get used as screens — pick the fund with the lowest tracking difference, avoid the one trading at a discount — and both uses are shakier than they appear. Tracking difference is largely a restatement of costs, which you can read directly. A fund lagging by 20 basis points with a 18-basis-point charge is doing fine; one lagging by 20 with a 5-basis-point charge is doing badly. The raw gap tells you little without the charge alongside it, and the charge is disclosed. A small tracking difference is not automatically good. A fund can reduce tracking difference by taking securities lending revenue, which introduces counterparty and collateral considerations, or by using synthetic replication, which tracks very closely and introduces a different risk entirely. Tighter tracking is sometimes bought with risk rather than skill. Historical figures may not persist — index changes, fund size changes, and market conditions change, and a past tracking record is a description rather than a forecast. Premium and discount are properties of a moment, not of a fund — the figure you see is historical, and what matters is the gap when you transact. And most importantly: both figures are second-order. The choice of index determines the great majority of a fund's outcome; the wrapper and the tracking quality determine a small residual. An investor agonising over 4 basis points of tracking difference while choosing between a global index and a single-country index is optimising the wrong variable by two orders of magnitude — the point the comparison article made about sequencing. These figures are worth understanding, worth checking, and not worth deciding on alone.