Key takeaways
- In Kacperczyk, Van Nieuwerburgh and Veldkamp's 1980 to 2005 sample, the managers their model flagged as skilled had 1.7 fewer years of experience than everyone else, at a p-value of 0.003.
- Pastor, Stambaugh and Taylor found each extra year of a fund's age cut gross benchmark-adjusted return by 15 basis points a year, across 3,126 US equity funds from 1979 to 2011.
- Berk and van Binsbergen's most-skilled decile of managers beat the least-skilled decile 61.57% of the time at a seven-year horizon and 54.69% at ten years.
- Morningstar's scorecard methodology, revised in 2019, puts the average portfolio manager tenure at five years, half the ten-year horizon over which manager skill has been shown to persist.
- Fidelity Contrafund's William Danoff has run the fund since 1990 and beat the S&P 500 by 2.03 points a year over the ten years to 31 December 2025.
Fund manager tenure is a sample size, not a performance signal
You're looking at a fund factsheet. It says the manager has been in the seat since 2011. You want to know whether that's a reason to buy the fund.
On the evidence, the tenure figure isn't telling you about the manager's ability. It's telling you how much data you have on that ability. A longer record makes any estimate of skill more reliable, and that's worth real money to anyone trying to pick a fund. It doesn't make the skill itself bigger. In the studies below, wherever experience or age enters the regression directly, the sign points the wrong way for the tenure story. The one exception is an effect its own authors decline to lean on.
That distinction runs through everything below. It also explains why the number is on the factsheet in the first place, why the industry's five-year average is short against the horizons this research uses, and why the longest-running counter-example here ends in December.
The number on the factsheet exists because the SEC put it there in 2004
Tenure isn't on the page because anyone proved it predicts returns. It's there because it's mandatory. In a rule adopted on 27 August 2004 and effective from 1 October 2004, the SEC amended Forms N-1A and N-2 so that funds must state "the name, title, length of service, and business experience of each member of a portfolio management team."
The rule was about accountability, not forecasting. A fund that names its manager and dates their appointment can be held to that record. But once a number is printed next to a manager's name, readers treat it as a quality score, and nothing in the rule says it is one.
Morningstar does use it as an input. Its Scorecard Methodology, effective from January 2017 and revised in May 2019, requires the manager tenure data point for the longest-tenured portfolio manager before an active equity fund can be scored at all. The same document is candid about the raw material: "portfolio managers currently have an average tenure of only five years." Morningstar's stated reason for caring is not that experienced managers earn more. It's that "funds run by experienced managers are less likely to encounter abrupt manager turnover, thereby mitigating manager risk." That's an argument about stability, and it's a different claim from an argument about returns.
The skilled managers in the largest test had less experience, not more
Kacperczyk, Van Nieuwerburgh and Veldkamp built a skill measure that separates stock picking from market timing, and lets the mix shift with the business cycle. Their NBER working paper, dated November 2011, runs on monthly data from 1980 to 2005. Their indicator flags a manager as a skilled stock picker or not. Once they'd sorted managers into the two groups, they looked at who those people were.
The flagged-skilled group averaged 26.45 years of experience. The rest averaged 28.14. That's a gap of 1.69 years in the wrong direction for the tenure story, and it isn't noise: the p-value is 0.003. In the authors' own summary, skilled fund managers "are 2.6% more likely to have an MBA, are one year younger, and have 1.7 fewer years of experience."
Two things stop this being a knockout blow. The effect is tiny, roughly 6% of a career, so nobody should read it as evidence that inexperience helps. And it's a cross-sectional comparison across managers, not a record of what happened to any one manager as they aged. Still, if portfolio manager experience were the input people assume it is, this is the sample where you'd expect to see it, and the sign is negative.
Fund age and manager tenure are different numbers, and only one of them has been measured properly
The cleanest long-run evidence isn't about managers at all. Pastor, Stambaugh and Taylor's NBER working paper, dated February 2014, covers 3,126 actively managed domestic equity-only US mutual funds between 1979 and 2011. They regressed benchmark-adjusted gross returns on fund age with fund fixed effects, so the comparison is a fund against its own younger self.
Performance falls as funds get older. The age fixed effects "decline in an approximately linear fashion from 37 bp per month at age one to zero at age 12, after which they are roughly flat." The linear version of the same regression puts it at 1.23 basis points a month, "or 15 bp per year," for each extra year. Sorting funds into age buckets gives the same answer: the youngest funds, aged three years or less, beat funds older than ten years by 7.2 basis points a month, "or 0.9% per year."
Here's the part that matters for tenure, and it's the part most write-ups drop. The age effect vanishes when the authors control for the size of the active management industry. Their reading is that new funds arrive more skilled than the incumbents, and the incumbents' edge erodes as capable competition piles in. Nothing about the manager got worse. The water got more crowded.
The same paper contains the only positive experience result in this piece. When they let skill vary with age to allow for learning on the job, skill grows by 1.51 basis points a month, "or 18 bp per year," for each additional year. They explicitly decline to lean on it, because the coefficient isn't clearly significant. Note the arithmetic anyway: 18 basis points of learning against 15 of erosion is close to a wash, and the erosion is the better-measured of the two.
A long record does carry information, and here is exactly how much
The strongest case for a long manager track record comes from Berk and van Binsbergen's NBER working paper, dated June 2012 and revised in August 2012. Rather than measuring alpha, they measure the dollar value a manager adds, across 5,974 funds using Vanguard index funds as the benchmark. The average manager adds "$140,000 per month" in year-2000 dollars, about $2 million a year. The distribution is brutal: the median manager lost "$20,000/month and only 43% of managers had positive estimated value added."
Their persistence test is where tenure enters, though they never use the word. They rank managers on a Skill Ratio, which the paper defines as essentially the t-statistic of "the value added estimate measured over the entire history of the fund until that time." Read that definition slowly. It is a t-statistic over the whole record, so the length of the record is an input to the ranking, not a separate question. A manager gets into the top decile partly by being good and partly by having been observed for long enough to prove it.
So how well does that ranking travel? The chart shows the share of the time the top decile went on to beat the bottom decile, out of sample, at horizons from three to ten years. The best result is 61.57% at seven years. The worst is 54.21% at nine. At ten years it's 54.69%.
That's a genuine edge, and it's statistically real at almost every horizon. It is also, at nine and ten years, about four to five points better than a coin. Skill exists, it persists, and knowing which decile a manager sits in still leaves you wrong nearly half the time.
Thirty-six years of tenure, and it ends this December
If you wanted to build the best possible case for fund manager tenure, you'd build it out of William Danoff. Fidelity Contrafund's prospectus, filed with the SEC on 24 February 2026, states that Danoff "has managed the fund since 1990." For the ten years to 31 December 2025, the fund returned 16.85% a year before taxes against 14.82% for the S&P 500. That's 2.03 percentage points a year, over a decade, from a manager three and a half decades into the job.
Two things complicate the story, and both are in the same filing.
The first is tax. The prospectus also reports the fund's ten-year return after taxes on distributions and the sale of shares: 13.63%. The index figure of 14.82% reflects no deduction for fees, expenses or taxes, so the two aren't strictly comparable. But a US taxable investor who bought this fund, held it for the decade and sold, kept a number 1.19 points below the untaxed index. Three and a half decades of tenure, and the after-tax margin on a liquidation basis isn't there.
The second is that the record is ending. The same prospectus says it "is expected that Mr. Danoff will retire effective on or about December 31, 2026." Two co-managers were added in 2025. Whatever the 1990 date told you about the fund, it stops telling you anything in a few months, and it stopped being the whole answer the day the co-managers arrived.
There's a survivorship problem underneath this too. Danoff is famous partly because his record survived to be famous. The FCA found that "worse performing funds were more likely to be closed or merged into better performing funds", which is the mechanism that removes the long tenures that went badly. You don't get to read their factsheets.
What happens to a fund when a long-tenured manager leaves
Clare, Sapuric and Todorovic built a database of UK fund manager changes from April 2002 to December 2005 and ran an event study around each one. Their finding on the winners is uncomfortable for the tenure story. The top 10% of funds before a manager change "continue to outperform, but only for a very short period until performance declines." Their explanation is that the departing manager's positions keep working for a while, and then stop.
The losers behave differently. Being in the bottom 10% before the change "makes little difference to their subsequent performance, so that underperformance persists at least for the following 12 months." A new name in the manager slot didn't fix a bad fund inside a year.
That asymmetry matches what the FCA found across the whole UK market. Its Asset Management Market Study final report, MS15/2.3, published in June 2017, concluded that "there is little evidence of persistence in outperformance in academic literature, but there is some evidence of persistent underperformance." The same report found UK active funds beat their benchmarks before charges and underperformed "on an annualised basis by around 60 basis points (bps)" after them, and identified "around £109bn in 'active' funds that closely mirror the market" while charging significantly more than passive alternatives. For a fund in that £109bn, the manager's tenure is describing how long someone has been running a tracker at active prices. What a fund actually holds tells you more than who has been holding it, and that is the sort of thing a fund factsheet shows you line by line.
Where this evidence runs out
Start with the mismatch at the centre of it. Almost all the good long-run data measures fund age, not manager tenure. Pastor, Stambaugh and Taylor measure age "by the number of years since the fund's first offer date", which is a property of the fund, not of whoever is running it. Their 15 basis points a year is a fund-age number. Treating it as a manager-tenure number is an inference, not a finding, and this piece is drawing that inference on your behalf.
The samples are old and they're mostly American. Kacperczyk and co-authors stop in 2005. Berk and van Binsbergen's data ends in 2011. Pastor, Stambaugh and Taylor stop in 2011 too. None of them covers anything after 2011. Since the industry-size mechanism they identify makes the answer depend on what the active industry did next, that gap matters more than a fifteen-year gap usually would.
The non-US evidence is thinner and it points the same way, which is worth noting and not worth overweighting. Mamatzakis and Xu's study of Chinese equity funds, posted to the MPRA archive in 2017 and covering 2005 to 2013, found that tenure "has a negative impact on equity fund performance at the 5% significance level," with a coefficient of 0.006 on the Sharpe ratio. Their tenure variable averages 2.68 years and tops out at 8. A range that narrow can't tell you what a 20-year record is worth, and a market that young can't tell you much about London or New York.
The UK event study has its own limits: a window of under four years, ending in 2005, on a single market. And every one of these studies measures what happened, not what was going to happen. None of it forecasts.
One more limit is structural. When a fund is run by a committee, "manager tenure" is a modelling choice rather than a fact. Morningstar carries two separate factors for it: one measuring "the number of years that the longest-tenured portfolio manager has been managing the fund", the other "the average number of years that the current portfolio management team has been managing the fund". On Contrafund, which now lists three co-managers, those two numbers differ by decades.
What would change the conclusion
A study that measures tenure the way funds report it. Manager start dates have been mandatory disclosure since 2004, so there are now more than twenty years of machine-readable tenure data sitting in SEC filings. Nobody in the sources above used it. A fund-age proxy is what these authors used; a test built on the filings themselves could reverse the sign.
A period when the industry stopped growing. The age effect disappeared once industry size was controlled for. If the active industry stops expanding, the crowding mechanism implies the erosion that made older funds look worse should weaken, and long records should start looking better without any manager getting more skilled.
Evidence that tenure predicts survival rather than return. This is Morningstar's actual claim, and its methodology document asserts it without publishing a figure behind it. If long-tenured managers really are less likely to leave abruptly, tenure would be a useful predictor of continuity even with a return coefficient of zero. For an investor holding a fund for two decades, that's arguably the more valuable forecast, and it's a completely different question from the one the tenure number gets asked.
The figure to watch on a factsheet isn't the start date. It's the length of the record relative to what you'd need to distinguish this manager from a lucky one, which the Berk and van Binsbergen numbers put somewhere north of ten years and still short of certain. Most managers never get there. Comparing what a fund holds against what its benchmark holds, as with fund overlap, or checking what past winners did next, as with active fund underperformance data, answers more in an afternoon than a start date does in a decade.