An ETF tracks an index by holding a portfolio designed to move in lockstep with that index, then relying on a behind-the-scenes trading mechanism to keep its market price aligned with the value of those holdings. There are three ways a fund can build the portfolio: buy every security in the index at its exact weight (full replication), buy a carefully chosen subset that behaves like the whole (representative sampling), or skip owning the securities entirely and use a swap contract with a bank to deliver the index’s return (synthetic replication). Which method a fund uses depends on the size and liquidity of the index it targets, and each choice carries its own cost and risk profile that shapes what you actually earn as an investor.
Full Replication: Owning the Whole Index
Full replication is the most literal approach. The fund buys every security in the target index at the same weight the index assigns. If a company represents 6% of the index’s total value, the fund puts 6% of its money into that stock. Most major indexes use market-capitalization weighting, so larger companies automatically get heavier weights based on their total market value.1Vanguard Professional. What to Consider When Choosing Between Index-Weighting Approaches The result is a near-perfect mirror of the benchmark.
This method works best when the index contains a manageable number of liquid securities. Tracking 500 large U.S. stocks is straightforward because every name trades actively and is easy to buy in size. The asset-weighted average expense ratio for index equity ETFs sat at 0.14% in 2024, with the cheapest large-cap funds charging as little as 0.03%. Fully replicated funds tend to cluster at the low end of that range because they don’t need heavy modeling or frequent optimization. They just hold the list.
Full replication funds can also earn extra income by lending portfolio securities to short sellers and other borrowers. The borrower pays a fee, and the fund keeps most of it. For broad-market ETFs, lending revenue is modest, but for funds holding harder-to-borrow stocks it can meaningfully offset the expense ratio and even improve net tracking performance.
Representative Sampling: Holding a Subset That Behaves Like the Whole
When an index contains thousands of securities or reaches into illiquid corners of the market, buying everything becomes impractical. A total bond market index might include more than 10,000 individual bonds, many of which trade infrequently and carry wide bid-ask spreads. Instead of chasing every last holding, the manager selects a subset that statistically behaves like the full index, matching it on characteristics such as sector exposure, duration, credit quality, and market-cap distribution.
Optimization software drives the selection. It identifies which combination of holdings will minimize the gap between the fund’s returns and the index’s returns, balancing competing constraints: match the index’s risk profile as closely as possible while keeping transaction costs and turnover low. Doing this well takes real skill. A well-run sampling fund tracking a broad bond index can stay within a few basis points of its benchmark; a poorly run one drifts noticeably.
The tradeoff is higher costs and slightly less precise tracking than full replication. Academic research has found that sampling funds tend to carry expense ratios roughly 0.14 percentage points higher than their fully replicated peers tracking the same benchmark, and they trade substantially more often. That extra trading generates friction that full replication avoids. Still, for indexes where full replication would mean owning thousands of thinly traded securities, sampling is the only workable option.
Every fund must describe its tracking methodology in the prospectus it files with the SEC on Form N-1A. That filing explains whether the fund fully replicates or samples, what characteristics the optimizer targets, and how the manager expects those choices to affect tracking performance. It tells you exactly how much discretion the manager has in choosing holdings.
Synthetic Replication: Getting the Return Without Owning the Securities
Synthetic replication skips owning the underlying securities entirely. Instead, the fund enters a total return swap with a counterparty, usually a large investment bank. The bank agrees to pay the fund the exact return of the target index, including dividends, and the fund pays the bank a fee.2Invesco. Considering Swap-Based ETFs The legal framework for these contracts is the ISDA Master Agreement, which standardizes how obligations, defaults, and terminations are handled across the derivatives industry.3International Swaps and Derivatives Association. 2002 ISDA Master Agreement Protocol
This approach can produce tighter tracking than physical methods because the bank contractually guarantees the index return. It also opens exposure to markets where directly holding the underlying assets would be difficult or expensive, such as certain commodity indexes or markets with foreign ownership restrictions. Swap fees vary but typically add a cost layer on top of the fund’s management fee.
Counterparty Risk
The vulnerability is obvious: the fund’s return depends on a bank’s promise. If the counterparty defaults, the fund is left holding whatever collateral was posted rather than the index return it was promised. Synthetic ETFs are typically over-collateralized to buffer against this, meaning the counterparty posts assets worth more than the swap’s current value. But collateral tends to be liquidated exactly when markets are stressed, which is the same environment that makes defaults more likely, so the protection is imperfect.
Federal regulation addresses this through SEC Rule 18f-4, which requires any fund using derivatives to adopt a formal risk management program covering counterparty exposure, set quantitative risk guidelines, conduct stress testing, and report material risks directly to the fund’s board.4U.S. Securities and Exchange Commission. Use of Derivatives by Registered Investment Companies and Business Development Companies – A Small Entity Compliance Guide
Where Synthetic ETFs Show Up
Synthetic replication is far more prevalent in European markets than in the United States. Most U.S.-listed ETFs use physical replication, either full or sampled, and the synthetic approach here is largely confined to funds tracking commodities or niche strategies where direct ownership is impractical. European UCITS-regulated funds have a longer history with swap-based structures, so investors encountering synthetic ETFs will most often find them on non-U.S. exchanges.
The Creation and Redemption Process
Holding the right securities is only half the job. The other half is keeping the ETF’s market price aligned with the value of those holdings, and that job falls to the creation-and-redemption mechanism. It’s arguably the most important piece of ETF architecture, and it runs through authorized participants: large broker-dealers with a contractual relationship to the ETF issuer. They can create or redeem shares in large blocks called creation units, typically at least 25,000 shares at a time.5Schwab Funds. Understanding the ETF Creation and Redemption Mechanism
When an ETF trades at a premium to the value of its holdings, an authorized participant has an incentive to buy the underlying securities on the open market, deliver them to the ETF issuer, and receive newly created ETF shares in return. Selling those new shares on the exchange captures the price difference, and the added supply pushes the ETF’s price back down toward net asset value. The reverse happens when the ETF trades at a discount: the participant redeems ETF shares for the underlying securities and sells them, reducing ETF share supply and nudging the price upward.
SEC Rule 6c-11 standardized this framework, requiring ETFs to disclose their full portfolio holdings daily on their website and establishing uniform conditions under which funds can operate without individual exemptive orders from the Commission.6U.S. Securities and Exchange Commission. SEC Adopts New Rule to Modernize Regulation of Exchange-Traded Funds Before this rule, each ETF operated under its own exemptive order with varying provisions.7Securities and Exchange Commission. Exchange-Traded Funds – Conformed to Federal Register Version
Primary Market vs. Secondary Market
Individual investors never touch the creation-redemption process directly. When you buy ETF shares through a brokerage account, you’re trading on the secondary market, buying from another investor on the exchange just like buying any stock. The primary market exists exclusively between authorized participants and the ETF issuer, and transactions there involve millions of dollars worth of shares at a time. The two layers work together. The secondary market provides the minute-to-minute trading liquidity you experience. The primary market acts as a pressure valve that corrects pricing misalignments between the ETF and its holdings.
How Well the Fund Actually Tracks
Two metrics tell you how well a fund is doing its job, and they measure different things.
Tracking difference is the total return gap between the fund and its benchmark over a specific period. If the S&P 500 returned 10.00% last year and your ETF returned 9.86%, the tracking difference is negative 0.14%. That number tells you what the fund’s replication cost you in real terms. For large, well-run S&P 500 ETFs, the tracking difference tends to land very close to the expense ratio, within a basis point or two, which is exactly what you’d expect.
Tracking error is the consistency of that gap over time, measured as the standard deviation of the periodic return differences. A fund with 0.02% tracking error barely wobbles around its benchmark. A fund with 0.50% tracking error delivers a return that bounces unpredictably relative to the index, even if the average tracking difference looks fine. For investors who care about predictability, which is most of the reason people buy index funds in the first place, tracking error matters as much as tracking difference.
What Causes the Gap
Several operational factors drive the wedge between a fund’s return and the index’s return:
- Expense ratio. The most predictable drag. A fund charging 0.14% per year will, all else equal, underperform its index by roughly that amount.
- Cash drag. Indexes are fully invested at all times. Real funds hold small cash buffers for redemptions and pending dividend reinvestment, and that cash earns less than the index return.
- Transaction costs. When the index adds or removes securities, the fund must trade. Commissions, bid-ask spreads, and market impact chip away at returns.
- Sampling mismatch. Funds using representative sampling never perfectly replicate every movement of the full index. The optimization gets close, but not identical.
- Fair-value pricing. Funds holding international securities sometimes adjust prices to account for time-zone gaps between when foreign markets close and when the ETF’s NAV is calculated. These adjustments can create small, temporary differences.
- Securities lending income. This works in the opposite direction. Lending revenue offsets some of the expense ratio drag, occasionally pushing tracking difference slightly positive relative to what fees alone would predict.
Checking both metrics before buying an ETF takes about thirty seconds on any fund screener. A large tracking difference relative to the expense ratio signals something is going wrong operationally. High tracking error means the fund’s returns are less predictable than its peers. Either one is a reason to look at a competing fund tracking the same index.
Costs the Expense Ratio Doesn’t Show
The expense ratio gets all the attention, but it’s not the only cost of owning an ETF. Two others affect what you actually earn.
Bid-Ask Spreads
Every time you buy or sell an ETF on the exchange, you pay the bid-ask spread, the gap between the highest price a buyer is willing to pay and the lowest price a seller will accept. For heavily traded funds tracking large-cap indexes, this spread is often a penny or two per share. For funds tracking small-cap stocks, international markets, or niche sectors, spreads widen because the underlying securities themselves are less liquid. A market maker setting prices for an emerging-market bond ETF faces real risk in assembling and hedging that basket, and the spread reflects that cost.
Spreads also widen when volatility spikes or when the underlying market is closed. A U.S.-listed ETF tracking Japanese stocks keeps trading after the Tokyo exchange closes for the day, but the market maker is pricing shares without live quotes on the underlying holdings. That uncertainty gets priced into a wider spread.
Premiums and Discounts to NAV
An ETF’s market price doesn’t always equal the net asset value of its holdings. When demand is strong, the price can drift above NAV, creating a premium. When selling pressure dominates, it can dip below, creating a discount. The creation-and-redemption process corrects these gaps, but the correction isn’t instant. Delays in accessing the underlying market, especially for international or fixed-income ETFs, can leave premiums or discounts lingering for hours or even days.
During periods of extreme volatility, these deviations can become significant. If you’re buying an ETF at a 0.50% premium to NAV, you’re paying half a percent more than the underlying securities are worth before the fund even begins tracking. Most fund websites publish the current premium or discount daily, and checking it before trading prevents overpaying.
The Hidden Cost of Rebalancing on Schedule
Indexes are not fixed lists. They update periodically to reflect changes in the market. Rebalancing adjusts the weight of existing securities, and reconstitution adds companies that newly qualify and removes those that no longer meet the criteria. MSCI, for instance, conducts a full quarterly review of its investable universe at each rebalance.8MSCI. Quarterly Index Review When those changes happen, every fund tracking the index has to trade in the same direction on the same day.
That’s where real money gets lost. Research covering S&P 500 rebalances from 2019 through 2023 found that stocks being added saw their prices spike during the closing auction on rebalance day, then reverse downward by the following morning. Index funds systematically bought at inflated prices. Over that five-year period, the cumulative cost of these reversals for S&P 500 trackers amounted to roughly $150 million in value left on the table. Trading volume during closing auctions on rebalance days ran eight to twenty-seven times higher than normal, depending on the index.
Active managers and hedge funds know this pattern and position themselves ahead of announced index changes, buying stocks expected to be added and selling those expected to be removed, which amplifies the price pressure that index funds face. This cost doesn’t show up in the expense ratio and doesn’t appear in the tracking difference calculation, because the index itself incorporates the same closing prices. It’s invisible to most investors, but it’s real, and it’s one reason some fund managers deliberately trade a day or two before or after the official rebalance date when their mandate allows the flexibility.