An index fund is built to match an index, yet its reported return almost never matches exactly. The gap has identifiable causes, and most of them are structural rather than errors.
Fees are deducted before the return is reported
The index itself is a calculation. It owns nothing, pays no custodian, files no paperwork, and carries none of the running costs that a real portfolio has to cover.
A fund does carry them, and the expense ratio is taken from assets steadily through the year. That deduction alone puts the fund a predictable fraction behind the thing it copies.
This part of the difference is the most stable of all. It appears in rising and falling markets alike, and it compounds along with everything else held in the account.
Cash inside the fund behaves nothing like the index
Investors buy and sell fund shares continuously, so a fund holds cash between receiving money and deploying it, and again while raising money to meet redemptions.
The index assumes full investment at every moment. When markets rise, idle cash drags the fund below the benchmark; when markets fall, the same cash cushions the decline instead.
Managers reduce the effect using futures that stand in for the index until the money is invested properly. The substitution is close rather than exact, which leaves a small residue.
Index changes have to be traded in the real market
When an index adds or drops a company, the fund must buy or sell it. The index makes that switch on paper, at a stated price, with no cost attached.
Real trades move prices, especially when many funds following the same index must transact on the same day, in the same direction, on a schedule everyone can see.
The cost of that crowding falls on the fund and its holders. It surfaces as another sliver of underperformance around each scheduled reconstitution of the underlying index.
Full replication is not always possible
Broad indexes contain thousands of names, some of them thinly traded or restricted to local investors. Holding every one in exact proportion would cost more than the mismatch it prevents.
Funds therefore sample, holding a subset chosen so the portfolio's characteristics resemble the index. Sector weights, size and other exposures are matched even where individual holdings are not.
Sampling tracks well in calm conditions and less well when a handful of omitted names move sharply. The residual shows up as tracking error rather than as a constant drag.
Securities lending pushes in the other direction
Funds lend out shares to borrowers who need them, mainly short sellers, and receive a fee plus collateral in return. That income belongs to the fund and its shareholders.
Lending revenue can offset part of the expense ratio, and in some markets it offsets most of it. A fund can occasionally finish a year slightly ahead of its index.
The programme carries its own risks, chiefly counterparty and collateral risk, and the revenue varies with borrowing demand. It is a real contributor to tracking difference, not an accounting quirk.