Why a Physical ETF Does Not Have to Hold Every Index Constituent
Representative sampling is a form of physical replication that uses selected assets to capture an index's characteristics. Look beyond holdings counts to exposures and actual tracking results.
Physical replication means directly holding underlying securities and includes both full replication and representative sampling. Sampling aims to approximate an index's overall risk characteristics using only some constituents. A smaller holdings count alone does not mean active management or failed index tracking.
Full replication is not always the simplest approach
When an index has few constituents and they trade readily, holding them at index weights is easy to understand. Buying and maintaining small positions in thousands of bonds or infrequently traded securities, however, incurs costs. A security's inclusion in an index does not mean the desired quantity can be purchased in the market. Bond indexes in particular can include many issues with similar characteristics. In such cases, using selected securities to represent the risks can reduce the number of trades.
Look at the risks represented, not just the sample size
For equities, sampling considerations can include country, sector, and company size; for bonds, maturity, duration, credit rating, and issuer weights. Matching risk characteristics differs from selecting a handful of securities with similar past returns. After listing the index's and ETF's holdings counts, place their main exposure weights side by side. If the sample is tilted toward a particular sector or part of the interest-rate curve, tracking may look good in normal conditions but diverge during a shock. A short period of calm performance does not establish representativeness.
Evaluate the trade-off between cost savings and tracking risk
Even holding every constituent does not produce perfect index performance because of cash, expenses, and distribution timing. Sampling reduces trading burdens but requires management of differences caused by sample selection. That is why products tracking the same index should be compared using both costs and tracking results across different market conditions. Also check whether the prospectus-permitted methods and actual holdings approach are fixed to a single method. Rather than inferring replication coverage from the product name, read the management-method section alongside the latest holdings. Checking whether holdings are dated before or after an index rebalance helps prevent a simple timing difference from being mistaken for a sampling problem.
Worked example
Hypothetical example: A bond index allocates 40% to short-term bonds and 60% to long-term bonds, with respective durations of 2 and 7 years. Its simple weighted duration is 5 years. A sampling ETF allocating 70% to short-term and 30% to long-term bonds instead has a duration of 3.5 years. Even a seemingly sufficient holdings count does not ensure the same interest-rate sensitivity. Matching the 5-year duration would still not establish identical credit risk or sensitivity to yield changes at different maturities.
What to check
- Have you recognized sampling as a form of physical replication?
- Have you compared the risk characteristics of index weights and actual holdings?
- Have you reviewed tracking results in both calm and stressed markets?
- Have you considered trading-cost savings alongside sampling risk?
Sources and primary materials
This is educational material about research methods. For an actual company decision, verify the filings and primary sources for the relevant period.