Editor’s note: this is general educational information about how index funds and their benchmarks are measured, not investment advice. It is based on the regulator guidance, rulebook text and index methodology listed at the end.
Analysis: the gap is manufactured by the rules on both sides
Put the two rulebooks side by side and the residual becomes predictable in origin, if not in size. The index reinvests a dividend on the ex date, at the index price, with no cash ever changing hands. A fund holding the same shares receives that dividend later, in cash, and buys at whatever price the market shows when it buys. The index moves its constituents after the close on a single named Friday each quarter. A fund has to transact into that change, in size, in the same window as every other fund tracking the same benchmark. Neither difference is an error in the ordinary sense. Each is the consequence of one document describing a calculation and another describing a portfolio.
The permissions in COLL point the same way. A regime that required exact holdings would force a fund to buy the smallest and least liquid constituents at any cost, so COLL 5.2.32 allows a result consistent with replication instead, and COLL 5.2.31 relaxes the spread limit to 20%, and in exceptional market conditions to 35% for one body, precisely so that a concentrated index remains trackable. Those choices reduce trading friction and, by design, introduce holdings that differ from the benchmark. The disclosure regime then handles what the flexibility creates: state the anticipated tracking error up front, state the realised figure at period end, and explain the divergence.
What none of these documents provides is a benchmark for what counts as a good tracking error, and none of them sets a numerical limit. That judgement is left to the reader of the annual report. The checkable items are the ones the rules name: which replication method the prospectus discloses, whether the report explains the divergence between anticipated and realised tracking error rather than only printing a number, whether the annual tracking difference is disclosed alongside it, and whether the fund’s largest single holding sits near the 20% relief that COLL 5.2.31 provides.
What the documents say
Tracking error is one of the few fund statistics with a supervisory definition rather than a marketing one. ESMA’s guidelines on exchange traded funds and other UCITS issues define it as the volatility of the difference between the return of the index-tracking UCITS and the return of the index or indices tracked. That wording settles two things at once. The number measures a difference in returns, and it measures the volatility of that difference, not its size.
What the disclosure regime asks for
The guidelines put tracking error in the prospectus before the fund has any performance to report. The prospectus of an index-tracking UCITS should include a clear description of the indices and their underlying components, information on how the index will be tracked, for example whether the fund will follow a full or sample based physical replication model or a synthetic replication, and the implications of that choice for investors in terms of exposure to the underlying index and counterparty risk. It should also include information on the anticipated level of tracking error in normal market conditions, and a description of the factors likely to affect the fund’s ability to track the index, the guidelines naming transaction costs, small illiquid components and dividend reinvestment. The replication method must also be summarised in the key investor information document.
Reporting closes the loop after the fact. The annual and half-yearly reports of an index-tracking UCITS should state the size of the tracking error at the end of the period under review. The annual report should explain any divergence between the anticipated and the realised tracking error for the period, and should separately disclose and explain the annual tracking difference between the performance of the fund and the performance of the index tracked.
Those are two different disclosures about two different quantities, and the guidelines treat them that way. A fund that ends the year close to its index but got there through choppy daily deviations has a small tracking difference and a larger tracking error. A fund that lags by a steady margin every day has the reverse profile. Reading either number alone answers only half the question, which is why the annual report is asked to carry both, and to explain the gap between the tracking error the prospectus anticipated and the one the fund produced.
Why exact replication is not the rule
UK rules do not require an index fund to hold the index. The FCA’s Collective Investment Schemes sourcebook states in COLL 5.2.32 that, in the case of a UCITS scheme replicating an index, the scheme property need not consist of the exact composition and weighting of the underlying in the relevant index in cases where the scheme’s investment objective is to achieve a result consistent with the replication of an index rather than an exact replication. Replication is defined by reference to the composition of the underlying assets of the index, including the use of techniques and instruments permitted for efficient portfolio management.
The concentration rules also bend for index funds, which matters when a benchmark is top heavy. COLL 5.2.31 allows a UCITS scheme to invest up to 20% in value of the scheme property in shares and debentures issued by the same body, notwithstanding the general spread rule, where the stated investment policy is to replicate a relevant index. That limit can be raised for a particular scheme up to 35% in value of the scheme property, but only in respect of one body and only where justified by exceptional market conditions.
The index itself has to qualify. Under COLL 5.2.33 a relevant index is one whose composition is sufficiently diversified, which represents an adequate benchmark for the market it refers to, and which is published in an appropriate manner. An index is an adequate benchmark if its provider uses a recognised methodology that generally does not result in the exclusion of a major issuer of the market to which it refers. It is published appropriately if it is accessible to the public and the provider is independent of the index-replicating scheme, which does not preclude the two being in the same group provided effective conflict of interest arrangements are in place. Where a scheme replicates an index, the FCA also points authorised fund managers to the additional disclosure requirements that apply to marketing communications.
The benchmark is a rulebook, not a portfolio
An index moves on the index provider’s calendar, and a fund has to move with it. The FTSE UK Index Series is reviewed quarterly, in March, June, September and December, with a full review of the FTSE All-Share, FTSE All-Small and FTSE 350 Yield Indices in June. Reviews are based on data in the FTSE UK Monitored List at the end of day on the Tuesday before the first Friday of the review month, and constituent changes are implemented after the close of business on the third Friday of the review month, following the expiry of the ICE Futures Europe futures and options contracts.
Weightings shift for reasons that have nothing to do with prices. Constituents of the FTSE UK Index Series are adjusted for free float and foreign ownership limits, and a security needs a minimum free float of 10% to be eligible, with a narrow route in for a new security whose free float sits above 5% because of restrictions that will lapse. FTSE Russell’s own illustration is a new issue with a calculated free float of 9%, restricted by a 7% holding by a sovereign wealth fund subject to a 6-month lock-in, where the minimum criteria are deemed met because the free float rises to 16% on expiry of the lock-ins within 12 months from the first day of trading. A foreign ownership limit can bind instead: the ground rules describe a non-UK incorporated company with a calculated free float of 62% and a lower foreign ownership limit being included with an investability weight of 49%.
Income is modelled rather than received. In the FTSE UK Index Series all dividends are applied as at the ex-div date, and ordinary cash dividends are reinvested across the index on the dividend ex date. Special cash dividends are treated differently: the stock price is adjusted to deduct the dividend amount before the open on the ex date, and the special dividend is not included in the total return index calculation, unless the company has paid one in a recurring cycle on more than three consecutive occasions, in which case FTSE Russell will normally treat further such distributions as ordinary.
