Key takeaways
- Chen, Hong, Huang and Kubik put the fund size penalty at 65 to 96 basis points a year on gross returns, across 3,439 US funds from 1962 to 1999.
- Yan's 1993 to 2002 sample found the smallest fund quintile beat the largest by 17 basis points a month before fees, and 14 basis points after them.
- It's a liquidity constraint. Among the least liquid portfolios small funds beat large ones by 35 basis points a month, and among the most liquid the gap vanished.
- The one bias-free estimate puts a $100 million increase in assets under management at 2.5 basis points a year, and it isn't statistically significant.
- Over the 10 years to June 2025 the average dollar in active funds beat the average active fund in 17 of the 20 Morningstar categories examined.
Does a fund get worse as it gets bigger? It's the warning that follows every fund that has a good run: the money piles in, and the manager who once slipped quietly into a small position now moves the price when they trade.
The effect is real and it's smaller than the warning implies. The studies that find it put it between about 65 basis points and 2% a year before fees, concentrated almost entirely in funds that hold illiquid stocks. The single best-identified estimate finds nothing at all. And in the returns investors actually received, the money sitting in the biggest active funds has done better than the average active fund, because big funds are cheap funds.
Fund size and performance: the penalty tops out near 2% a year, in gross returns
The paper everyone cites is Joseph Chen, Harrison Hong, Ming Huang and Jeffrey Kubik, "Does Fund Size Erode Mutual Fund Performance? The Role of Liquidity and Organization", published in the American Economic Review in December 2004. They used 3,439 distinct funds and 27,431 fund years from 1962 to 1999.
Their result: a two-standard-deviation shock to the log of a fund's assets under management moved next month's return by 5.4 to 7.7 basis points, "or about 65 to 96 basis points annually", depending on the benchmark. After fees the figures were only slightly smaller.
That sounds tiny until you set it against what these funds were doing anyway. The same paper notes that its funds "on average underperform the market portfolio by about 96 basis points after fees and expenses". So the spread between a big fund and a small one was roughly the size of the entire shortfall against the index.
Xuemin Yan re-ran the question on 1993 to 2002 data in the Journal of Financial and Quantitative Analysis in 2008, using stock transaction data to measure how liquid each fund's holdings actually were. His smallest size quintile averaged $37.61 million; his largest held over $4 billion. Using the Carhart four-factor model, the smallest quintile beat the largest by 17 basis points a month on gross returns and 14 on net. A two-standard-deviation increase in log assets cost "almost 18 basis points per month, which translates to over 2% per year" in his sample, against "less than 1% per year" in Chen's.
The mechanism is liquidity, not a manager going soft
Both papers point at the same cause, and it isn't effort. Yan sorted funds by the liquidity of what they held. Among funds holding the widest-spread stocks, the smallest beat the largest by 35 basis points a month. Among the three most liquid quintiles, he found "no evidence that small funds outperform large funds". Chen's team reached the same conclusion from the other direction: outside small-cap funds, "size does not significantly affect performance".
What a growing fund does with the money explains the rest. Yan's largest quintile held 162 stocks; his smallest held 72. Lubos Pastor, Robert Stambaugh and Lucian Taylor, studying 2,789 active US equity funds from 1979 through 2014, found the pattern holds industry-wide: "Larger funds are cheaper. Larger and cheaper funds trade less and are less active, based on our novel measure of activeness."
So a fund doesn't get stupider as it grows. It gets more diversified, trades less, and drifts closer to the index it's benchmarked against. That's a real cost if you're paying an active fee for it, and it's a different complaint from the one people usually make. It's also why a fund's stated fund factsheet holdings count tells you more about drift than its asset total does.
The one study designed to remove the bias finds almost nothing
Here's the strongest objection to everything above, and it's serious. Fund size isn't assigned at random. Good funds attract money, so size and skill are correlated, and a plain regression of returns on size measures both at once.
Pastor, Stambaugh and Taylor tackled this in a 2014 National Bureau of Economic Research working paper using recursive demeaning, a procedure built to strip out that bias. Their ordinary regressions found the usual negative relationship. The bias-free version did not. A $100 million increase in fund size "depresses performance by 0.0022% per month, or 2.5 bp per year", and in their longer sample "only 1.3 bp per year". Neither was statistically significant.
Jonathan Reuter and Eric Zitzewitz attacked it differently, using the fact that a fund just above a Morningstar star-rating cutoff takes in far more money than a near-identical fund just below it. That's as close to a randomised trial as fund data gets. Across Morningstar's US universe from December 1996 to August 2009, "the average Wald estimate is 0.028 versus an average OLS estimate of -0.002" for the effect of size on returns. Correcting measured performance persistence for scale moved it "from 0.091 to 0.078", a change they couldn't distinguish from zero.
The academic dispute is live, and the sceptics lost the last round
In 2021 Adams, Hayunga and Mansi argued that data errors produced a small number of influential observations behind both the industry-level and the fund-level findings. Pastor, Stambaugh, Taylor and Min Zhu answered on 13 July 2021 in "Diseconomies of Scale in Active Management: Robust Evidence". Their finding: "We reject constant returns to scale even after dropping 25% of the most extreme return observations."
Where that leaves it: diseconomies of scale at the fund level survive robustness testing, and their measured size stays small. The largest estimate in the papers here is Yan's, at a little over 2% a year, and that is the gap between funds two standard deviations apart in log assets.
Industry size does most of the damage that fund size gets blamed for
The same 2014 paper found something the fund-level debate keeps missing. Measure the whole active mutual fund industry as a share of US stock market capitalisation, and it starts "at 2.4% in January 1979, peaks at 18.6% in July 2008, and finishes at 16.8% in December 2011". Run performance against that instead of against a single fund's assets and the relationship is strong: a one percentage point rise in industry size cost the average fund "0.0326% per month, or almost 40 bp per year".
Fund by fund, 62% of the individual slope estimates were negative. And the authors show that managers didn't get worse while this happened. Average measured skill rose "from 24 basis points (bp) per month in 1979 to 42 bp per month in 2011". Funds under three years old beat funds over ten by 0.9% a year on gross benchmark-adjusted returns, and that age gap disappeared once industry size was controlled for.
Read that carefully, because it reframes the question. Your fund's returns fading over a decade is consistent with your fund never having changed at all. What changed was how many skilled competitors were doing the same trade.
The money in the biggest funds has done better than the average fund
Now the awkward part for the diseconomies story. Morningstar's US Active/Passive Barometer for midyear 2025 covers 9,204 funds holding roughly $24 trillion, about 68% of the US fund market at the end of June 2025. It reports every category twice: once giving each fund equal weight, once weighting by assets, which loads the average onto the biggest funds.
Over the 10 years to June 2025, "the average dollar invested in active funds outperformed the average active fund in 17 of the 20 categories examined". The chart above plots that gap in percentage points for seven of those categories. In US large blend, the asset-weighted active return was 11.9% a year against 10.9% equal-weighted, a full point in favour of size. Only in a minority of categories, Europe stock and corporate bond among them, did the big funds trail.
UK data says something similar and quieter. In the SPIVA Europe Scorecard for year-end 2024, sterling-denominated UK equity funds returned 5.33% a year equal-weighted over 10 years and 5.29% asset-weighted, against 5.99% for the S&P United Kingdom BMI. Over three years the asset-weighted average returned 0.91% against -0.27% equal-weighted. Size cost UK investors 4 basis points a year over the decade, and earned them more than a point over three years.
These two bodies of evidence don't contradict each other, and it's worth being precise about why. The academic studies compare funds of different size after holding fees, age, turnover and style constant. An asset-weighted average holds nothing constant. It mostly measures where the money went, and the money went to the cheap funds. Morningstar found 27% of active funds in the cheapest quintile beat their average passive peer over the decade, against 15% of the priciest. The scale penalty is real and the fee discount that arrives with scale is bigger.
Active fund persistence is thin at every horizon, whatever a fund's size
None of this makes picking a small fund a way in. The S&P U.S. Persistence Scorecard for year-end 2024 tracks every active US domestic equity fund from a starting rank and asks how many hold it. Below, the share still in that group after each further year, starting from December 2020.
| Group at December 2020 | After 1 year | After 2 | After 3 | After 4 |
|---|---|---|---|---|
| Top half, all domestic equity | 32.91% | 7.93% | 5.09% | 4.21% |
| Top half, large-cap | 39.70% | 5.45% | 2.73% | 2.42% |
| Top half, small-cap | 21.91% | 9.56% | 4.38% | 2.79% |
| Top quartile, all domestic equity | 5.27% | 0.00% | 0.00% | 0.00% |
Random chance would keep about half a top-half group there after one year, roughly 25% after two, and 6.25% after four. Every row in the table sits below those baselines at every horizon. The large-cap row runs 39.70%, then 5.45%, then 2.42%.
The small-cap row is the interesting one for this argument. Small-cap is the category where the scale penalty is largest and best documented, so if fund size were doing much of the work you'd expect the most persistence there. It has the lowest first-year figure in the table, at 21.91%. The table doesn't sort funds by size, so this is an observation rather than a test, but it isn't the direction a strong size effect would predict.
Survivorship pushes the same way. Of the fourth-quartile US domestic equity funds ranked on 2014 to 2019 performance, 25% were merged or liquidated within five years, against 7% of top-quartile funds. The funds that disappear are disproportionately the small ones, which is exactly the population a size-based rule sends you towards. Anyone using past ranking as a shortlist is running into the same wall that performance chasing runs into, and the broader base rates sit in the SPIVA scorecard.
Scale economies go to the manager, scale diseconomies go to you
The Financial Conduct Authority put the asymmetry in writing in its Asset Management Market Study final report, MS15/2.3, in June 2017. On the cost side: "At a fund level, we have found that asset managers tend to benefit from economies of scale and these do not seem to be passed on fully to investors."
On the return side, the same document notes that larger funds "can experience diseconomies of scale but these tend to affect funds' capacity to invest, and so lead to lower returns to investors, rather than create higher costs for asset managers".
Those two sentences are the whole trade laid out by a regulator. Growth makes a fund cheaper to run, and the saving is shared at the manager's discretion. Growth makes a fund harder to run, and that cost lands on the return. Whether the arrangement leaves you ahead depends on how much of the cost saving reaches your expense ratio, which is where the compounding effect of fund fees actually shows up.
What this evidence cannot tell you
Almost all of it is American. Chen's sample runs to 1999, Yan's to 2002, Pastor's to 2014. The SPIVA Europe scorecard is the only UK evidence here, and it's a fund average, not a size sort. Nobody has published the equivalent size regressions on UK funds at that depth.
Every size sort compares different funds at one moment. It doesn't follow one fund from £200 million to £5 billion and watch what happened, which is the question a holder actually has. Reuter and Zitzewitz built their whole design around that gap and still came back with confidence intervals too wide to settle it.
There's a deeper problem underneath. If money flows to skill until expected excess return is competed away, then in equilibrium size and future returns are uncorrelated by construction. A null result would then mean the market has already priced the diseconomies, not that they don't exist. Reuter and Zitzewitz say as much, and it's why their null is weaker evidence than it looks.
These are sample averages over particular decades, not forecasts. The 1962 to 1999 window and the 1993 to 2002 window produced estimates that differ by a factor of two on the same question. And none of it covers investment trusts, or funds that hard-close before they reach the sizes being measured.
What would change the conclusion
If market liquidity deteriorated, the whole effect grows. The penalty is a trading-cost penalty, and Yan's result is that it exists only where spreads are wide. A market structure that widened spreads on mid-caps would push the size effect out of the small-cap corner it currently lives in.
If fee competition stopped, the offset disappears. The reason asset-weighted returns beat equal-weighted ones is that scale currently comes bundled with a discount. That bundle is a feature of the last two decades, not a law.
If the active industry shrank materially from the 16.8% of US market capitalisation it reached in December 2011, the Pastor mechanism predicts surviving funds' gross alpha would recover, and the fund-level question would matter more than it does now.
The figure worth tracking isn't the fund's asset total on its own. It's the liquidity of what the fund holds set against how much of it the fund owns, because that ratio is where every estimate above actually comes from. A fund holding £5 billion of FTSE 100 shares and a fund holding £5 billion of UK small-caps are not the same animal, and the research has never once said they were.