Key takeaways
- Active share measures the fraction of a portfolio that differs from its benchmark; Cremers and Petajisto labelled non-index funds below 60%, with tracking error under 6%, closet indexers.
- ESMA screened 1,251 EU equity funds over 2012 to 2014 and flagged between 5% and 15% as potential closet indexers, depending on the thresholds applied.
- The FCA estimated in June 2017 that around £109bn sat in UK active funds that closely mirror the market while charging significantly more than passive funds.
- A fund charging 1.00% with 40% active share, against a 0.24% index alternative, collects an effective 1.9% a year on the slice that actually differs from the index.
- In the 1990 to 2003 record, the least active fund quintile underperformed by 1.41% to 1.76% a year after fees, while the most active quintile beat its benchmarks by 1.39%.
The question, and the short answer
How much of your active fund is actually active? You're paying an active fee, so it's a fair thing to want to know. For a meaningful minority of funds, the measured answer is: much less than the fee implies.
The measure is called active share, and it's simply the percentage of a portfolio that differs from its benchmark index. A pure index fund scores near 0%. A portfolio with no overlap at all scores 100%. Martijn Cremers and Antti Petajisto computed it for 2,650 US equity funds across 1980 to 2003. They found that funds with low active share, between 20% and 60%, held about 30% of all mutual fund assets by 2003, compared with almost zero in the 1980s. Funds in that band mostly hold their index, charge an active fee, and describe themselves as active. The authors named the practice closet indexing, and the name stuck.
Two European regulators later went looking for it with the same yardstick and found it. This piece walks through what they measured, what the practice costs in fee terms, and what the performance record says about funds that hug their benchmark. The arithmetic in the middle section is the part worth remembering.
Active share counts the part of your fund that isn't the index
Every portfolio benchmarked to an index can be split into 2 pieces: a position that replicates the index, and a set of active bets layered on top. Active share is the size of that second piece. Formally it's half the sum of the absolute differences between the fund's weights and the index weights, taken across every stock. Informally, it's the fraction of the portfolio that is different from the index.
Cremers and Petajisto's insight, in a working paper dated 7 August 2006 and later published in the Review of Financial Studies, was that active share and tracking error measure different things and belong together. Tracking error, the volatility of a fund's return relative to its benchmark, is the industry's traditional yardstick, and index fund analysis leans on the related tracking difference vs tracking error distinction. But a diversified stock picker can hold a genuinely different portfolio while showing modest tracking error, because stock-level bets diversify. A fund that is low on both dimensions, yet still claims to be active, is the interesting case. The authors drew the line conservatively: a non-index fund with active share below 60% and tracking error below 6% got the closet indexer label. Their canonical example was Fidelity Magellan as of 2002, then a $62bn fund.
The trend behind the label was stark. The share of US equity fund assets with active share below 60% rose from 1.5% in 1980 to 40.7% in 2003. Over the same period, the average active share of non-index large-cap funds fell from about 80% to about 60%. Closet indexing was not a fringe habit; it was where a large slice of the industry's assets had quietly moved.
A 1.00% fee on a 40% active fund is 1.9% on the part doing the work
Here's the arithmetic the label exists to expose. In the Cremers and Petajisto data for 2002, funds with active share between 30% and 60% charged a value-weighted expense ratio of about 1% a year. The purest index funds charged 0.24%. An investor in the low-active-share fund is paying roughly 0.76 percentage points a year more than the index alternative, and the only thing that premium buys is the minority of the portfolio that differs from the index.
Spread that premium over the slice it pays for and the effective price of the active piece emerges. At 40% active share, 0.76% divided by 0.40 is an effective 1.9% a year on the active slice. At 30%, it's 2.53%. At 20%, it's 3.80%. Only at 100% active share does the effective rate fall to the headline premium of 0.76%. The chart above runs the numbers across the range. The mechanism compounds the same way ordinary fund fees do, but at 2 to 5 times the visible rate, because most of the portfolio is doing work an index fund does for a fraction of the price.
The study's fee tables explain why the label is about disclosure rather than about fees as such. Expense ratios barely varied across the active share spectrum for non-index funds; the most active funds charged a value-weighted 1.47%, against roughly 1% for the 30% to 60% band. In the authors' words, active funds charge similar fees regardless of their active share, so the ones that hug the benchmark are most likely doing it without acknowledgement to their investors.
How much closet indexing regulators found when they went looking
The academic finding became a supervisory question. In February 2016 ESMA, the EU's markets regulator, published the results of a screen of UCITS equity funds. It started from more than 2,600 funds with over €50mn under management, a track record predating 2005 and management fees above 0.65%, and retrieved usable data for 1,251 of them over 2012 to 2014. Using the academic thresholds, active share below 60% with tracking error below 4%, 15% of the sample was flagged as potentially closet indexing. Tightening to active share below 50%, tracking error below 3% and an R-squared above 0.95 still left 5%. ESMA noted that in their prospectuses, the vast majority of the flagged funds described their management approach as active.
The FCA reached a similar conclusion for the UK in its Asset Management Market Study final report of June 2017. Its estimate: around £109bn in active funds that closely mirror the market while being significantly more expensive than passive funds. The report also noted that charges for passive funds had fallen on average while charges for active funds had remained broadly constant. That combination is why an index hugging fund's fee premium had widened rather than narrowed.
The Central Bank of Ireland went next. In July 2019 it published what it described as its largest data driven thematic review of the funds industry to date, covering all 2,550 Irish authorised UCITS classified as actively managed as at March 2018. Its press release published no count of closet trackers. What it reported instead was arguably worse. Investors were not always given sufficient or accurate information about the strategy. And some funds had a target outperformance against the index lower than the fee charged, so that even a fund hitting the top of its own projections would leave those share classes behind the benchmark.
The performance record: paying full fare for a seat that barely moves
Closet indexing would matter less if low-conviction funds performed adequately. The record says they perform like an index fund minus their fee, which is exactly what the structure predicts. In Cremers and Petajisto's 1990 to 2003 sample, the lowest active share quintile of non-index funds underperformed its benchmarks by 1.41% to 1.76% a year after fees, depending on the risk adjustment. Before fees, the same funds roughly matched their benchmarks. Zero skill, minus expenses.
The highest active share quintile, by contrast, beat its benchmarks by 1.39% a year net of fees in the same sample, though the authors are careful to note that this point estimate falls just short of statistical significance. Truly active funds at least gave themselves the chance to earn the fee. The closet trackers structurally couldn't.
Whether paying for high conviction is a good bet is a separate question, and the broader scorecard evidence is not kind to active management as a group. The SPIVA scorecard comparison, as of 30 June 2025, put the share of US large-cap funds underperforming the S&P 500 at 65% for the full year 2024, at 86.91% over 5 years, 85.98% over 10, and 91.03% over 20. And picking the winners in advance runs into the persistence problem. The S&P Persistence Scorecard for year-end 2024 tracks how long outperformers stay on top, by horizon:
- 1 year on: of the 176 top-quartile US large-cap funds of 2022, 0% were still top quartile at the end of 2023, against 25% if rankings were random.
- 2 years on: still 0% of that cohort, against an expected 6.25% by chance.
- 5-year run: of the 162 top-quartile large-cap funds of December 2020, 6.79% remained top quartile after 1 year and none after 2, 3 or 4 more years.
- Top half, 5 straight years: 2.42% of 330 large-cap funds managed it, versus the 6.25% a coin flip predicts.
- Beating the index repeatedly: of the 374 large-cap funds that beat the S&P 500 in 2022, 2.14% did so again in each of the next 2 years.
That table frames the closet indexing question honestly. Even genuinely active funds rarely repeat. A fund with 40% active share is charging you active fees for a diluted version of a bet that mostly fails at full strength.
The case against active share, made by the industry and by the data
The measure has serious critics, and the FCA's consultation recorded their strongest arguments. Respondents argued that active share and tracking error are not accurate proxies for how active a fund is, that mechanical thresholds risk mislabelling genuinely active funds, and that tracking error is not a determinant of future outcomes. The FCA partly conceded the point: it agreed that neither tracking error nor active share should be the sole basis for identifying a partly active fund. It repeated all the same that over £100bn appeared to closely follow the market at active prices.
The objections have substance. A fund benchmarked against the wrong index can score a misleading active share. ESMA itself ran a looser second screen because, in its words, that set of metrics could be more indicative in member states with relatively small equity markets, where portfolios sit structurally closer to a concentrated national index. And the performance case for high active share is weaker than it first looks. Cremers and Petajisto themselves report that among the largest 40% of funds, even the most active did not add value after fees and transaction costs; the outperformance sat mainly in smaller funds. They also found that half of all active positions cancel out across funds, so a randomly chosen active fund delivers a useful active share of no more than about 30% toward any aggregate result.
None of that rescues the closet tracker. Every criticism above is an argument about whether high active share earns its fee. A fund at 35% active share charging 1% fails on the arithmetic no matter which side of that debate wins, because the index fund replicating most of its portfolio exists and costs a quarter of a point.
What this evidence cannot tell you
The foundational numbers are old and American. The 2,650-fund sample ends in 2003, and its fee levels, 1% versus 0.24%, date from 2002. Index fund charges in particular have fallen since, on the FCA's own account, which changes the inputs to the arithmetic without changing its structure. ESMA's screen is a snapshot of 2012 to 2014, and its 5% to 15% range is explicitly a set of statistical flags, not findings of misconduct. ESMA itself called the results a first step requiring fund-by-fund follow-up. The FCA's £109bn is an estimate based on tracking error, a method its own respondents attacked. The Central Bank of Ireland's published findings concerned disclosure and governance rather than any count of closet trackers.
There's also a measurement gap a reader can't easily close. Active share is computed from full portfolio holdings, and funds don't disclose it in any consistent way. ESMA observed during its review that where such metrics are disclosed, there is no consistency in their use. Without that number, an investor can observe fees precisely and activity only roughly, through tracking error, overlap with the index, or a factsheet's own benchmark language. The asymmetry is the heart of the problem: the fee is certain, the activity it buys is not.
What would change the conclusion
The arithmetic depends on the fee premium, so fee compression attacks it directly. If a closet tracker charged 0.35% rather than 1%, the effective cost of its active slice at 40% active share would fall from 1.9% to about 0.3% a year, and the label would lose most of its sting. Cheap and slightly active is a defensible product; expensive and slightly active is the problem.
Mandatory disclosure would change it from the other side. Both ESMA and the FCA chose disclosure remedies over hard thresholds, and a world where every fund prints its active share in the KIID is one where the 30% funds either cut fees, raise conviction, or lose assets. The measure of whether that world arrived is simple: whether you can find the number without computing it yourself.
And a robust out-of-sample repeat of the high active share result, at scale, in the post-2003 record, would turn the measure from a cost screen into a selection tool. The original authors found the outperformance concentrated in smaller funds and short of statistical significance at the quintile level. So the honest reading today is narrower: active share reliably tells you what you're paying for management, and only unreliably whether the management is worth paying for. The number to watch, for any fund you hold, is the effective fee on its active slice, not the fee on the label.